US 2210028 A
Description (OCR text may contain errors)
Aug. 6, 1940.
W. H. DOHERTY AMPLIFIER Original Filed April 1. 193a s'sheetssheet 1 U INVEN7DR By W H DOHERT) A T TORNEY Aug; 6, 1940. w. H. DOHERTY I AMPL IF IER Original Filed April 1, 1936 5 Sheets-Sheet 2 Q 2 c L m Ml N. w. 2 U C m M 7 W m. M Q F 7 INVENTOR W H. DOHERTV Arm/war 9 w. H. DOHERTY 2210,028
I AMPLIFIER Original Filed April 1, 1936 3 Sheets-Sheet 3 FIG. /0
TUNER FIG. I?
I AMPLIFIER TUNER ll d :3 r INVENTOR W H. DOHERTY ATTORNEY 3 I 555 la F19 S TUNER I Patented Aug. 6, 1940 UNITED PATENT omen Application April 1, 1936, Serial No. 72,147
Renewe'd pctober 18', 1938 47 Claims. (c1. 179- 171) This invention relates to vacuum tube ampli-' fiers and more particularly to power amplifiers for'signal'modulated carrier waves.
The principal objects" of the invention are to secure increased efficiency in the operation of power amplifiers and to provide for the maintenance of this increased efficiency under all conditions of modulation.
Power amplifiers for modulated carrier waves ciencies as high as from 65 to '75 per cent with,
constant high frequency input voltages.
When the input voltage is modulated themaximum efficiency, with the circuit arrangements heretofore used, is obtained only at the-peak of the modulation cycle, the efficiencies at other instants in the cycle being less in proportion tothe. voltage amplitude. The average efficiency. over a modulation cycle, for any .given degree of modulation, is therefore always considerably less than the maximum efliciency obtainable under steady conditions, its' value ranging from about 37 per cent forzeromodulation to about 52 per cent for 100 per cent modulation.
By the present invention, the average efficiency may be increased to greater than 60 per cent for zero modulation and may be maintained at this high level for all degrees of modulation. Where amplifiers of large power are used, as in radio broadcast transmitters, the annual power saving represented by the increased average efiiciency is verylarge, particularly since the average degreeof modulation over any extended operating period is only about 20 per cent.
The increased efiiciency is achieved in part by dividing the vacuum tubes of the amplifier into two groups, in one of which the tubes are con-; tinuously operative and in the other of which the-tubesare operative. only when the input voltage is greater than that of the unmodulated carrier. Since the tubes of the second group consume no power during one-half of a modulation cycle and reach their full power consumption only atthepeak of the modulation, the net. re-
sult is a reduction'of the average power required; Increased efficiency is obtained also by the use of coupling and driving circuits of novel character, the details of'which aredescribed here'-" inafter, whereby the tubes of the'first group are enabled to supply their proper share of the out-=- put power at constant output voltage duringthe intervals when the tubes of the second group are" also operative. This permits'the tubes of the first group to be operated attheir" maximum output voltage and'th'ereforeat their maximum emciency for all input voltages in excess of the normal carrier voltage. The special circuit 'con-" nections may be regarded as functioningto vary the effective loadimpedance into which the tubes of the first group operate, the impedan'ce dimin' ishing as the'impressed voltage increasesso that the power output increases even though the out put voltage, and therefor'e'the efiiciency,'rem'ains constant.
A characteristic feature of the invention is that'the'circuit coupling one-of the groups of tubes to the load'circuit is a circuit of the type having 'aphase constant of some *odd integral multiple 0 f"- v i n v radians Circuits of this type-have certain properties which accomplish the functions described above. Among these are the following: A voltage at one terminal of the circuit is associated with a definitecurrentat the other terminal irrespective of the terminating impedance; with a constant input voltage the output or load voltage is a linear function of the load impedance; and the input impedance of the'circuit is inversely proportional to the load impedance. .Because of this last characteristic, such circuits are sometimes referredto as impedance inverting networks.
The nature of the invention and its manner of operation will be more thoroughly comprehended from the following detailed description of a number of itsembodiments and from the attached drawings of which:
Figs. 1 and la are-schematicsillustrating the general circuit configurations of the invention;
Figs. 2 and 3 are diagrams explanatory of the operation of the invention;
Figs. 4, Sand 6 illustrate types of coupling networks used in the invention;
Fig. 7 illustrates properties of the coupling networks;
Fig. 8 shows a modified form of coupling network;
Figs. 9, 10 and 11 are detailed schematics of two specific embodiments of the invention; and
Fig. 12 discloses a modified form of the invention embodying feed-back stabilization.
Figs. 1 and la show in a generalized form the two fundamental species of the invention; Fig. 1, that in which the two amplifiers are connected in parallel to the load and Fig. 1a that in which they are connected in series. Figs. 9 to 11 show specific embodiments of the species of Fig. 1. This parallel connected species which is illustrative of the fundamental principles of the invention will be described first.
The amplifying system, illustrated in Fig. 1, comprises two parallel transmission paths interconnecting a source S of amplitude modulated carrier waves and a resistive load R, each path including a vacuum tube amplifier and its associated input and output networks. For the sake of clarity the details of the networks and of the vacuum tube power supply circuits are omitted. The vacuum tube amplifiers are designated I and 2 and are shown as single tubes although parallel arrangements of tubes may be used. Networks 3 and 4 are phase shifting networks for controlling the relative phases of the voltages impressed on the amplifiers. Networks 5 V and 6 are tuned circuits for selectively transmitting the modulated carrier waves and are preferably broad-band wave filters having transmission bands substantially wider than the frequency range occupied by the carrier and its attendant side-bands. Networks 3 and 4 should also, preferably, have relatively wide transmission bands. B1 and B2 designate sources of negative grid bias voltages for the amplifiers.
Amplifier l is biased to about the plate current cut-off by the voltage of source B1. This amplifier is continuously operative for all amplitudes of the alternating voltage impressed on its grid. Amplifier 2 has a considerably greater negative bias on its grid, the bias voltage being such as to permit the space current to fiow only when the voltage of the impressed oscillations exceeds that of the unmodulated carrier.
Coupling network 5 in the output of the continuously operative amplifier is proportioned to have a phase constant which, in the general case, is an odd integral multiple of In practice it is preferable to have the phase constant of this network equal to since this requires a minimum of impedance elements, but under special conditions, for example, where strong suppression of harmonics is desired, more complex networks giving greater discrimination may be used. Network 6 in the output of the intermittently operative amplifier is proportioned to have a phase constant equal to an even integral multiple of at the carrier frequency. Preferably the phase constant of this network should differ from that of network 5 by and when the latter network has a phase constant equal to the phase constant of network 6 may be zero. That is, the output of the amplifier may be connected directly to the load.
Phase shifting networks 3 and 4 are proportioned to produce phase changes of the impressed voltages complementary to those produced by coupling networks 5 and 6 to insure that the final output currents will combine in phase with each other in the load impedance. Generally these networks may be simple well-known phase splitting circuits for producing voltages in quadrature, for exciting the grids of the respective tubes.
The principles underlying the operation of the invention will be developed from a consideration of the circuit of Fig. 2 which is representative of the output portion of the amplifier shown in Fig. 1 for the simplified case in which the network 5 has a phase constant The space path of amplifier l is represented by a resistance R1 with a wave source of voltage E1 connected in series and that of amplifier 2 by a corresponding resistance R2 and voltage E2. The voltages E1 andvEz difier in phase by the same amount as the respective input voltages differ from each other and are proportional to the input voltages in magnitude. Amplifier 2 is connecteddirectly to the load resistance R, the coupling network 6 of Fig. 1 having been reduced to a direct connection. The circuit coupling amplifier 2 to the load network 5 of Fig. l is represented by a T-network of pure reactances, the shunt and series reactances being of opposite sign and all three having the same magnitude X. This network is an equivalent of any reactance network having a phase constant of It is used here simply for theoretical purposes and is not intended to indicate a form that networks used in practice must have. Other structural forms are described later.
The equations for the mesh currents I1, I2 and I3 are the folowing:
Current I1 is the output current of amplifier I, I2 is the output current of amplifier 2, and I3 is the current delivered to the load from network 5. The total load current is the sum of I2 and I3.
As expressed by Equations 2 each of these currents is made up of two components, one proportional to the electromotive force E1 and the other to the electromotive force E2.
If the voltages Ei and E2 were in like phase,
each of the mesh currents and the load current would be made up of two quadrature components.
By proper phasing of thevoltages impressed on phase opposition. In accordance with the invention the relative phases of the inputs are such that E1 is in phase with jEe andhence that jE1 isin phase with E2. The two components of I1 defined by the first of Equations (2) are then-of like sign and are in phase with E1, the components of the load current are also of like sign and are in phase with E2, while those of I2 and I3 are of opposite sign. Uunder this condition; the output current of amplifier I is augmentedwhen amplifier 2 is delivering current and the power output from amplifier I for a given input voltage is therefore increased. In dealing with power amplifiers it is customary to describe the characteristics in terms of thejoutput voltages of the vacuum tubes rather than the electromotive forces generated in the space paths. This results in a simplification of the equations for the'mesh currents and in the case of the present invention serves also to bring out certain important operating characteristics. The values of the mesh currents in terms of the voltages V1 and V2 at the output terminals of the vacuum tubes may be obtained from Equations (2) by making resistances R1 and Re zero, the voltages E21 and E2 being then equal to the output voltages. Keeping in mind the condition stated above as to the relative phases of E1 and Ea'the following magnitude relationships are obtained:
From these equations may be obtained the.
further expressions for I1,
I1=v1 +1z (4) and Equations (4) and (5) show how the current I2 delivered from the intermittently operative amplifier 2 modifies the effective impedance into which the continuously operative amplifier I works, thereby permitting that amplifier to deliver an increased amount of power without a corresponding increase in its output voltage. Equation (5) also illustrates the impedance inverting characteristic of the quarter-phase network which couples amplifier I to'the load.
It will be evident that so long as amplifier 2 is delivering no current the impedance into which amplifier I works is equal to and is therefore inversely proportional to the actual load impedance. If the terminating impedance R is increased, the efiective impedance presented to the tube of the amplifier I will diminish and vice versa. This impedance inverting property is a well-known characteristic of quarter wave-length. lines and other reactance networksoperatihg at a frequency such as to make the phase constant equal to radians or any odd multiplethereof. In multiple section networkssuch as wave-filters the phase constant increases progressively with frequency in the transmission range and may pass through several odd multiples of hole at different frequencies." At each of such fre quencies the impedance inverting property is found. At these frequencies also the output current is independent of the value of the load impedance so long as the voltage at the input terminals is held constant. This is indicated by the expression for I3 in Equations (3) above.
- The power delivered by amplifier I is equal to a phase shift of any even multiple of radians.-
When amplifier 2 comes into operation the apparent resistance of the load is equal to and consequently increases as the current I2 increases; i. e., as the amplifier 2 supplies a greater portion of the load current. Because of the impedance inverting property of the coupling network the eifective' load impedance presented to amplifier I,is diminished as I2 increases and a larger power output is obtained from that amplifier without an increase in its output voltage. By proper adjustment of the amplifier input and biasing voltages, the rate at which the current Iz increases with the input voltage may be so regulated that the output voltage ofamplifier I remains constantduring the active half of the modulation period in which the amplifier 2 is operative. The additional power from amplifier I is thus obtained without reduction of its efficiency. I m V V The use of .the quarter-phase network in the output of amplifier I and the direct connection of the amplifier to the load whereby the above- 2 radians and the coupling from amplifier 2 maybe through a network having a phase shift of any even multiple of role radians. In the general case, therefore, the networks 5 and 6 in the output circuit may have phase shifts diifering'by any odd multiple of radians while the networks 3 and 4 in the input circuit have such complementary phase shifts as to make the total phase changes in the two amplifier paths equal or to bring the two load currents into phase. The essential feature is the use of the impedance inverting or quarter-phase network between one of the amplifiers only and the load and the compensation elsewhere of the phase displacement produced therein.
At the peak of the modulation cycle, when the total output power is a maximum, it is desirable that the load should be equally divided between the two amplifiers. This requires the proportioning of the values of the reactance X and the load resistance R with respect to each other as follows:
From Equation (3) the power outputs of the amplifiers are found to be ity of V111 and V2I2 requires that If the tubes of both amplifiers are of the same type, they will preferably be operated to deliver their maximum outputs at the same output voltages, in which case the required relationship between X and R becomes In special cases it may be desirable to operate the two amplifiers at different maximum voltages or to share the load between them in other proportions. In each case an appropriate relationship of the reactance to the resistance may be found and the basic proportions of the coupling thereby established.
For linear amplification it is necessary that the voltage across the load impedance should vary in direct proportion to the voltage impressed upon the system from the modulated wave source. In Fig. 1 the voltage from the input wave source is impressed on the system at terminals (1 a and the load voltageis measured at terminals b b. If these voltages are linearly related, the total load current, being directly proportional to the load voltage, will also vary linearly with the input voltage. In the simplified system illustrated in Fig. 2, where the output terminals of amplifier 2 are connected directly to the load, the load voltage is the same as the amplifier output voltage V2. Due to the character of the coupling circuits and to the phase relationships of the impressed voltages the magnitudes of the several currents and voltages in the amplifier output circuits are related in accordance with Equations In Fig. 3 the required variations of the output current and voltage amplitudes are plotted as functions of the source voltage, denoted by V0. In the figure the values of V0 are plotted as absc'issae, OArepresenting the unmodulated carrier amplitude and OB, which is twice 0A, representing the peak amplitude for 100 per cent modulation. Straight line OC represents the variation of voltage V2 required for linear amplification, this voltage being also the voltage at the load terminals. The total output current I2+I3 is also linear and is represented by straight line OD.
Output current I2 from amplifier 2 is zero for all input voltages lessthan OA and at higher voltages is representedby the line AE, rising to a value equal to half the load current at the modulation peak. Its variation from A to E may be assumed for the present to be linear. Cur- I rent I; is equal to the difference between the total load current and I2 and its required variation is therefore represented by broken line OFE. At input voltages greater than 0A the amplitude of Is remains constant. Current I1, which is proportional to V2 is represented by straight line OE. At the modulation peak it is equal to I2 on the assumption that the two amplifiers have the same output voltage and power at this point.
The output voltage V1 of amplifier l is directly proportional to Is and, since at the modulation The efficiency of the amplifier when operating under the output voltage and current conditions defined by the curves of Fig. 3 may be determined approximately as follows:
It will be assumed that the tubes of both amplifiers are similar and are operated with equal D. C. platevcltages. It will also be assumed that the carrier wave is modulated by a sinusoidal signal oscillation producing a percentage modulation equal to Hlilm, where m denotes the ratio of the variation of the carrier oscillation amplitude to the amplitude of the unmodulated carrier. Over the modulation cycle the instantaneous value of the power delivered to the load varies in proportion to the square of the load current. The average power depends upon the degree of modulation and has the value vary directly with the amplitude of current 11. Its average value therefore remains constant for all degrees of modulation and the plate power has the average value where W1 denotes the plate power, Ebv the D. C. plate voltage and Ibl the average plate current for zero modulation. The plate current in amplifier 2 flows only during alternate half periods'of the signal oscillation and occurs in the form of impulses which, due to the stronger grid bias, are
of shorter duration than half cycles of the carrier current. The amplitudes of the impulses follow the variations of the output current I2 but their duration is not constant and increases gradually with the amplitude. If the impulses were all of half wave duration the average value of the plate current would be equal to2mIb1:-1r, the value being expressed in termsof the plate cur-' rent'in amplifier I since at the modulation peak the plate currents in both tubes would be equal. However, because of the shorter duration of the current pulses, amplifier 2 will operate at a slightly higher efiiciency than amplifier I and the average value of the plate current will be somewhat less than that given above. To take account of this improvement in the eificiency the plate power consumed by amplifier 2 may be expressed by where W2 is the plate power and q a variable factor which ranges from about 0.7 for zero modulation to 0.93 for full modulation.
. From these expressions for the power output and power consumption, the average efliciency, denoted by 71m, throughout a modulation cycle is found to be 1-1-7 1 7| where 1 0 is the efficiency for zero modulation. Since amplifier I alone is operative at zero modulation and is delivering power at its maximum voltage, its efiiciency may be as high as 65 per cent or greater. As the degree of modulation increases theeificiency tends to'diminish slightly in accordance with Equation 12 and at full modulation may be about 94 per cent of its value for zero modulation.
While the characteristics illustrated in Fig. 3 have been developed for the simplified circuit of Fig. 2 it may be shown readily that they also obtain in the general case where the network 5 has a phase constant equal to an odd integral multiple of I and network 6 has a phase constant differing,
therefrom by That this is so maybe under: stood from the following, consideration. If'the two amplifiers were coupledydirectly to. thegload,-.;,
or through networks producing phase changes thatare even multiples of the output voltages of both amplifiers would nec-' essarily vary directly with the load voltage and hence, for linear amplification, would have to follow directly the variations of the voltage of the input source. Under such conditions it'would not be possible to operate the first amplifier at constant output voltage and at the same time secure linear over-all amplification.
Toobtain the required constancy of the output voltage V1 for outputs greater than the normal carrier output it is necessary that the signal voltage impressed on the grid of the first amplifier should vary in an appropriate manner. For
values up to the normalicarrier amplitude the grid voltage should be directly proportional to the voltage V0 of: the modulated wave source, but at larger amplitudes should increase more slowly than the wave source voltage. I have found that for most types of vacuum tubes the increase in the grid voltage between the normal carrier value and the value at the modulation peak should be about 30 per cent. The formof the variation of the impressed grid voltage is shown by the broken dotted line OKN in Fig. 3.
One method of securing the desired excitation characteristic is to adjust theinput circuit so that the grid of the first amplifier begins to draw current as soon as the input voltage amplitude exceeds the normal carrier value. When the grid becomes conductive, the' amplifier input impedance is reduced and a diminution of the efiective alternating grid voltage results. By the use of a grid leak, the direct component of the grid current may also be used to increase the negative bias on the'grid thereby efiecting a further reduction' of the output voltage. Both effects may be tude of the voltageimpressed on the grid from the. source. By appropriate adjustment'of these parameters the desired output characteristic may be obtained. v
With regard to the structural form of the phase shifting and selective networks, many schematic forms including the T-arrangement shown in Fig.2 may be used, but the equivalent schematic form shown in Fig. 4 is generally preferable in practice. This network, like that of Fig. 2, has a phase constant of when the reactances'are of the magnitudes and sign indicated. Changing the sign of each of the reactances changes the sign of vthe phase constant thereby producing a phase reversal of the output current and voltage. One specific form of network of the type shown in Fig, 4 is illustrated in Fig, 5, the structure of which comprises a seriesinductance L1, and twoequal shunt condensers of capaoity'Cz. This is a single section of a well-known form of low passfilter; At the when the reactances' of the series and shunt branches are of equal magnitude.
The frequency variation'of the phase constants of the networks of Figs. 5 and 6 are illustrated by the curves l and 8, respectively, of Fig. 7, which show their values as functions of the ratio of the variable frequency to the cut-off frequencies'of the filters. For the low-pass structure the phase constant is equal to at' a frequency equal to .707fc, ,fc being the cutoff frequency, and for the high-pass structure of Fig. 6 the phase constant has its desired Value at a frequency equal to 1.4l4fc.
As the frequency is varied through the transmission band the phase angles of the single section networks illustrated undergo changes of 1r radians or 180 degrees. If several sections are connected in tandem the phase constant at any frequency is increased in proportion to the number of sections. Since it is characteristic of most single section filters, whether low-pass, high pass, or band-pass that the phase angle changes by 180 degrees between the band limits; it follows that, in the case of the band-pass filter the rate of change of the phase shift will increaserapid- 1y as the width of the transmission band is reduced. If very narrow band filters are used for the selective circuits 3, 4, 5 and 6 the resultant phase shifts may vary very rapidly with frequency in the neighborhod of the operating frequency with the result that the side-band frequencies corresponding to the modulating signal will not all be subject to the desired phase shift of 90 degrees. For this reason it is preferable to use networks having band widths as great as possible consistent with an adequate degree of suppression of the harmonics of the carrier wave. In practice band widths of the order of kilocycles or greater have been found adequate for dealing with speech modulated waves. At carrier frequencies of 500 kilocycles and upwards this permits a substantial suppression of harmonies.
A form of band-pass network which has been found to be useful in practice is illustrated in Fig. 8 which corresponds to Fig. 5 with the addition of anti-resonant circuits LC in parallel with the shunt branches. If these anti-resonant circuits are tuned to the carrier frequency their effective reactance becomes infinite and the overall reactance relationship of the series and shunt branches at this frequency remains unchanged. The tuned circuits may be added at each end as illustrated or a single tuned circuit may be added at one end alone if desired. The network of Fig. 6 may also be modified in the same way without changing the over-all reactarice relation ship. In networks of this type the operating frequency lies substantially midway between the cut-off frequencies. Other well-known forms of broad-band filter networks may also be used, the
elements being appropriately proportioned so that the phase constant is equal to at the operating frequency. If desired also sev-.
eral sections may be connected in tandem thereby giving phase constants of any desired multiple of i The complete amplifier schematic of Fig. 9' illustrates one system of phase shifting and se-' lective networks and also circuit arrangements for providing the required input voltage characteristics. l is of the band-pass type shown in Fig.8 and network 4 in the input of amplifier 2 is of the same type, the phase constants of the two networks being of the same sign and magnitude so that the over-all phase changes in the two paths are equal. At their input ends the two paths are connected to adjustable taps on potentiometer 9 in the output circuit of the modulated wave source S. Biasing potentials are supplied to the amplifier grids from sources B1 and B2 through resistors ill and H, blocking condensers I2 and !3 being provided to isolate the grids from other,
parts of the circuit. Plate current is supplied to the amplifier tubes through choke coils l6 and justed so that at the normal carrier.ampiitude'the grid is just becoming noticeably conductive. As
the input voltage increases abovethis valuethe increased grid current flowing through resistor iii produces an increased negative bias .on the grid and a correspondingdiminution in the amplification with a resultant tendency toward reduction of the output voltage. At the same time the impedance presented by the input circuit diminishes due to the increased conductivity of the grid and the amplitude of the oscillations reaching the grid is likewise diminished. Both effects tend in the same direction and by proper adjustment may be made to maintain the required output voltage characteristic. The proper outputcurrent characteristic of amplifier Zis obtained by the coordinated adjustment of the voltage of biasing source B2 and of the amplitude of the input voltage by means of potentiometer 9 in the manner previously described. I In the modified form of the invention shown in Fig. 10 the phase shifting input network is in-' serted in front of amplifier 5 instead of in the input of amplifier 2. Equalization of the totalphase constants of the two paths is obtained by making the phase constants of thetwo networks equal in magnitude and opposite in sign. The output network 5 is of the same type as in Fig. 10
Network 5 in the'output of amplifier am -02s while the input network 3 corresponds to the type shown in .Fig. 6 with the addition of an antiresonant circuit at one end. The biasing voltage is fed to the grid through the adjacent coil of the network instead of through the terminating'resistor Ill. The grid bias is thus substantially unaffected by any fiow of grid current.
The output load is shown as an an coupled to the'amplifier througha tuning network I8, the combination being proportioned and tuned so that it presents to the amplifier a pure resistive impedance of the desired magnitude at the operating frequency. Network l8, may be a narrow band wave filter or other well-known form of coupling circuit and may be as sharply tuned as desired.
Control of the output voltage depends wholly upon the change of the tube input impedance when the grid becomes conductive. The theory of the operation of this method of control is as follows: Y
The input circuit may be regarded as being constituted by a net work of the type shownin Fig. 4 terminated at its output end by a resistance R2 which represents the resistor In in parallel with the grid to cathode space path, and supplied from a wave source of voltage E and effective internal resistance R1.
,At the operating frequency, for which the network has a phase constant equal to 5 the output voltage of the network denoted by'E'z: may be shown to be given by When" no grid current flows the ratio of the two voltages is determined by the substantially fixed resistances R1 and R2 and the fixed resistance X. When a substantial grid current fiows resistance R2 isreduced to a new lower value and the .ratio of the voltages is changed in such manner that the output voltage increases more slowly with increase of the-source voltage. By proper selection of the value of resistor the input voltage may be controlled to give the required output characteristic shown in Fig. 3. v
The voltage at the network input terminals, which is also the voltage impressed on the grid of amplifier 2 is given by E .RZ FX 4 where E1 denotes the input voltage. The reduction of R2 due to the flow of grid current when the source voltage rises above the normal carrier value, results in a corresponding increase in the voltage impressed upon amplifier 2. This voltage increase assists in bringing amplifier 2 into action more rapidly and in many cases may be sufl'icient to permit the two amplifier paths to be operated at the same signal input voltage.
Another modification of the invention is shown in Fig. 11 in which the control of the input voltages is effected'by regeneration. The two paths are coupled to the wave source S through a tuned transformer 23 and a feedback. circuit is con nected from the mid-point of the secondary of this transformer through a phase controlling network 2| 'to transformer in the plate circuit of amplifier tube 2. The feedback circuit is adjusted so that the voltage fed back is'in phasefopposition to the input voltageof amplifier I -and in tennai I 91' phase coincidence with that of amplifier 2 at the. operating frequency. Since no current flows in the plate circuit of amplifier 2 so long as the input voltage is less than the normal carrier value, the feedback becomes effective only when the input voltage exceeds this value and then it operates to reduce the grid voltage on tube I and to increase that on tube 2. By proper adjustment of the amount of feedback the desired output voltage characteristics can be. obtained. The primary winding of transformer 20 should preferably have a'rather low inductance so that its presence in the plate circuit will not seriously affect the phase of the output currents.
In Fig. 11 the two networks 3 and 5 are both of types giving positive phase shifts of the total phase change in the path being equal to 1r plus the phase change in the vacuum tube. To compensate this phase change the input voltage to the second path is reversed in phase by its connection to the secondary of tuned transformer 23 in the manner indicated in the drawing.
In the operation of the amplifiers of the invention itmay be foundthat some slight deviation from the ideal linear amplification characteristic occurs at voltages near the normal carrier value, this being due to an improper fit of the'characteristics of the two tubes andto lack through tuned coupling transformer 2'5. C'oupling between amplifiers 24 and 25 'is effected through tuned circuit 21 and condenser 28. The feedback is applied through conductor 29, connected to the high voltage side of the amplifier output, resistance 32, and a phase controlling filter consisting of inductance and condenser 3| to the input of amplifier 25. Since two amplification stages are used in this system a phase reversal of the feedback is necessaryto bring it into the proper degenerative relation. Thisis efi'ected by making the reactance of the coil 30 large relatively to that of condenser 31 "at the operating frequency. The amount of the feedback may be controlled by varying the inductance of the coil. e v v The degenerative feedback also makes it pose: ble to simplify the input circuit arrangements, since any lack of linearity due to improper relationship of the input voltage magnitudes in the two paths will be suppressed thereby. It becomes possible therefore to apply the same input voltage to each path as shown in the figure.
In Fig. 1a, which represents a modification of the circuit of Fig. l to the right of line XX, there is shown an alternative connection of the load impedance whereby it is supplied with cur-- rent from the two amplifiers in series instead of in parallel. The same operating characteristics are obtained and the same requirements on the input voltages apply as in the circuit of Fig. 1 Due to the series relationship of the load, however, it is necessary that the. coupling networks 3, i, and ii have their phase characteristics interchanged. Amplifier 2, which is intermittently excited, must be coupled to the load through i a network having a phase constant equal to or an odd multiple thereof while the output circuit of amplifier i should have a phase constant of zero or an even multiple of then the load would receive current only during the intervals that this amplifier is excited. By the modification of the phase constants of the coupling networks as indicated above the load can receive power at all instants.
What is claimed is: 1. An amplifying system comprising a source of modulated carrier waves, a load circuit, two parallel transmission paths connecting said source and said load, an amplifying device included in each of said paths, phase controlling means between said devices and said'source for producing a phase quadrature relationship be tween the waves impressed upon the input terminals of said devices from said source, and phase controlling means between the output terminals of said devices and said load for producing a compensating relative phase shift of the amplified oscillations whereby. they combine in additive phase in said load circuit and the impedance presented to the output of one of said amplifying devices varies inversely with the effective impedance of the load circuit.
2. An amplifying system for supplying amplitude modulated carrier waves comprising a wave' source, a load circuit, two parallel transmission paths connecting said source and said load, an amplifying device included in each of said paths, phase shifting means between at least one of said devices and said source for producing a phase quadrature relationship betweenthe waves im pressed upon the input terminals of said devices from said source, phase shifting means between at least one of said devices and said load for producing a compensating relative phase shift of the amplified oscillations whereby they combine in like phase in said load and whereby the impedance presented to the output of the lastmentioned one of said devices varies inversely with the terminating impedance of, the lastmen-' tioned phase shifting means, and biasing means for rendering one of said amplifying devices inoperative for input amplitudes less than a preassigned value.
3. An amplifying system for supplying amplitude modulated carrier waves comprising two amplifying devices, a Wave source, a load circuit, separate input paths coupling the input termi nals of said devices to said source, separate output paths coupling the output terminals of said devices to said load circuit, a reactance network in one of said output paths producing a phase shift of the amplified waves equal to 90 degrees at the frequency of said source, the phase shift in the second of said paths differing therefrom by 90 degrees whereby the impedance across the output terminals of that one of said amplifying devices connected to said one of said output paths varies inversely with the effective terminating impedance of said reactance network, and phase controlling means in said input paths producing,
radians at a preassigned operating frequency, coupling means in the other output path having a phase constant at said frequency equal to an even integral multiple of a phase constant equal to an odd integral multiple of radians at a preassigned operating frequency, coupling means in the other of said output paths having a phase constant at said frequency equal to an even integral multiple, including zero, of
radians, coupling means in said input paths having phase constants of such magnitudes at said preassigned frequency as to equalize the overall phase constants of the two paths between said common pairs of terminals, and biasing means for rendering one of said amplifying devices inoperative at input amplitudes less than a preassigned value.
6. An amplifying system comprising two amplifying devices, a pair of common input terminals, a pair of common output terminals, separate input paths coupling said devices to said input terminals, separate output paths coupling said devices to said output terminals, a reactance network in one of said output paths having a phase constant equal to radians at a preassigned carrier frequency, the other of said output paths having a phase constant equal to zero at said frequency, coupling means in said input paths having phase con stants of such magnitudes as to equalize the overall phase constants of the two paths between said common pairs of terminals, and biasing means for rendering one of said amplifyingdevices inoperative at input amplitudes less than a..pre-
. assigned value.
charge amplifiers, a load impedance, a 'wave source, a reactance network coupling one of said amplifiers to said load, said network having a phase constant substantially equal to radians at a preassignedoperating frequency, a separate coupling having substantially zero phase constant between the second of said amplifiers and said load, input circuits coupling said amplifiers and said source, means'in said input circuits for producing a quadrature phase relationship of the waves impressed upon said amplifiers,
and biasing means for rendering said second amplifier inoperative at input amplitudes less than a preassigned value.
8. An amplifying system for carrier waves of variable amplitude comprising a space discharge amplifier, a load impedance, a source of carrier waves, an impedance inverting reactance net work coupling said amplifier and said load, an input circuit coupling said amplifier and said ,source, an additional transmission path between said source and said load, said additional path and the path constituted bysaid amplifier and its associated coupling circuits having equal phase constants, andbiasing means for blocking said additional path to waves of less than a preassigned amplitude.
9. An amplifying system for supplying amplitude modulated carrier waves comprising a space discharge amplifier, a load imp dajnce, a source of carrier waves of preassigned frequencfa nd preassigned normal amplitude, a reactance network having a phase constant of 2 radians at said preassigned frequencypcoupling said amplifier and said load, an,. i nput circuit coupling said amplifier and said source ,an additional. transmission path between said source andsaid load, said additional, path and the path circuit, means for so exciting saidamplifying paths from said wave source that the outputs therefrom are in phase in said load circuit,
means for blockingsaid second amplifying path to waves of ,less ;than a predetermined normal amplitude, and ,m,eans for so controlling the op- 1 normal voltage, a reactance network having a phase constant of transmission path between said source andsaid load, a second space discharge amplifier included in said path and coupled to said load with zero phase shift, the two paths between said source and said load having substantially equal phase constants, biasing means for blocking said second amplifier to waves of less than said normal. voltage, and means for automatically controlling the amplitude of the waves impressed on said first amplifier in response to excesses ofsaid wave sourcefivoltage above said normal value whereby the output voltage of said first amplifier is maintained constant. a
12. An amplifying system in accordance with claim 11 in which the first-mentioned space discharge amplifier is biased substantiallyto the plate current cut-01f point.
, 13. An amplifying system in accordance with claim 11in which the said 'reactance network is proportioned with respect to thevalue of said load impedance to provide an equal division of the total output power between the said two amplifiers at a voltage of said wave source equal to twice the said normal value.
14. An amplifying system in accordance with claim 11 in which the said amplifiers are gridcontrolled vacuum tubes, the grid of' the firstmentioned amplifier being negatively biasedto substantially the plate current cut-off point, and the said normal wave source voltage being just suflicient to produce a positivegrid current in said first-mentioned amplifier.
15. Anamplifying systemin accordance with claim 11 in which the said automaticcontrolling means comprises a coupling, forproducing from the output circuit of said second amplifiera feedback voltage and supplying said feedback voltage to the input circuit of said first amplifier in opposition to. the; input voltage supplied thereto from said source.
16. An amplifying system in .accordanceuwith claim 8. and in, combination therewith degenerative feedback coupling means betweensaid load and said source.
17. In an amplifying system comprising a source of amplitude 'niodulated carrier waves, a load circuit, and a pair of amplifying devices included in separate transmission paths between said source and said load, an impedance inverting network coupling the output terminals of a first of said devices to said load, direct connections from the output of the second of said devices to said load, coupling circuits between said devices andsaid source having phase shift characteristics such that the output currents from said devices are in phase in said load circuit, means for blocking said second device to waves from said source of less than a predetermined normal amplitude, and means operative when the voltage from, said source exceeds said normal amplitude to .control the transmission between said, source and said first device whereby its output voltage is maintained constant.
18. In an amplifying system comprising a source of amplitude modulated waves of preassigned normal carrier voltage, an amplifying device, and. a load circuit, an impedance inverting network coupling the output terminals of said device and said load circuit, an input circuit coupling said device and said source, means responsive to excesses of the voltage of said source above said normal value for reducing the effective impedance at the output terminals of said device substantially in proportion to the magnitude of the excess voltage, and means simultaneously operative to control the transmission eificiency of said input circuit whereby the output voltage of said device is maintained constant.
19. In an amplifying system comprising a source of amplitude modulated waves of preassigned normal carrier voltage, an amplifying device, an input circuit coupling said device to said source, and an output circuit connected to the output terminals of said device, means responsive to voltages of said source above said normal value for reducing the effective impedance of said output circuit and means simultaneously operative to control the transmission efficiency of said input circuit whereby the output voltage of said device is maintained constant.
20. In an amplifying system, a source of electrical oscillations of varying amplitude, a load circuit, a first amplifier, circuits for connecting the input of said first amplifier to said source and the output thereof to said load circuit so that for oscillations from said source of amplitude below a predetermined level the output voltage of said amplifier is substantially linearly proportional to the amplitude of said oscillations and for oscillations from said source of a range of amplitudes above said predetermined level the output voltage of said amplifier is substantial- 1y constant, a second amplifier, and circuits for connecting the input of said second amplifier to said source and the output thereof to said load so that the output current from said second amplifier is substantially zero for oscillations below said predetermined level and substantially proportional to the amplitude of said oscillations for oscillations of said range of amplitudes above said predetermined level and so that the currents supplied by said amplifiers to said load are in phase.
21. A system according to claim in which at least one of said connecting circuits includes at least one network of reactive elements so arranged and proportioned as to have a phase constant of some odd integral multiple of 22. In an amplifying system, a source of electrical oscillations of varying amplitude, a load circuit, two amplifiers interconnecting said source and said load, means for so connecting and operating one of said amplifiers that for oscillations from said source below a predetermined amplitude the current supplied thereby to said load circuit is substantially linearly proportional to the amplitude of said oscillations and for a range of amplitudes above said predetermined amplitude the current supplied thereby to said load circuit is substantially constant, and means for so connecting and operating the other of said amplifiers that for oscillations from said source of amplitude below said predetermined amplitude substantially no current is supplied to said load circuit and for a range of amplitudes above said predetermined level the current supplied thereby to said load is substantially linearly proportional to the amplitude of the oscillations from said source and in phase with the load current supplied by said first amplifier.
23. A system in accordance with claim 22 in which said means include means for causing the load currents supplied to said load by said amplifiers to be equal for oscillations from said source of amplitude substantially twice said predetermined amplitude.
24. In an amplifying system, a source of amplitude modulated carrier oscillations, load circuit, a first vacuum tube amplifier, an input circuit for said first amplifier connected to said source, an output circuit for said first amplifier connected to said load circuit, said output circuit comprising reactance elements of such character and so arranged and proportioned that the impedance into which said amplifier works is inversely proportional to the impedance of said load circuit and of a value such that said amplifier operates at a high efficiency at the carrier amplitude of oscillations from said source, a second vacuum tube amplifier, an input circuit for said second amplifier connected to said source, an output circuit for said second amplifier connected to said load circuit, said output circuit being of such a character that the impedance into which said second amplifier works is directly proportional to the load impedance, and means in said input circuits for said first and second amplifiers for causing said second amplifier to be operative only for oscillations from said source greater than the carrieramplitude and for causing the currents supplied by said amplifiers to said load for higher amplitudes to be in phase and substantially equal for amplitudes of oscillations from said source twice the carrier amplitude.
25. A system in accordance with claim 24 in which said means comprises a phase shifting network.
26. A system in accordance with claim 24 in which said means comprises means for biasing the input of said second amplifier so that no plate current fiows for input oscillations below the carrier amplitude.
27. A system in accordance with claim 24 in which said means comprises a phase shifting network in the input circuit for said first amplifier.
28. A system according to claim 24 in which said means comprises a phase shifting network in the input circuit of said first amplifier, an input bias for said first amplifier and means for causing the input bias of said first amplifier to increase for input oscillations above the carrier amplitude.
29. In an amplifying system, a source of amplitude modulated carrier waves, a load circuit, a continuously operating amplifier, means for connecting the input of said amplifier to said source so that for voltages from said source below the carrier amplitude the voltage output of said amplifier is substantially proportional to the input voltage and for higher voltages from said source the output voltage of said amplifier is substantially constant, a circuit for connecting the output of said amplifier to said load comprising reactance elements so arranged and proportioned that the output current thereof is independent of the terminating impedance, a second amplifier, means for connecting said source to said second amplifier so that for voltages from said source below the carrier amplitude no output is produced in said amplifier and that for voltages from said source above the carrier amplitude the output voltageeof -"said :amplifier varies substantially proportionately with .the'r input voltage, and means. for connectingthe voutput of said second amplifier to said loadtso that the current supplied thereto from .thesecond amplifier is substantially directly proportional to the output voltage ofsaidvsecond-amplifier, saidmeans for connecting the inputsofsaid. continuously operatingamplifiera-nd said second amplifier to said source having suchiphasecon'stants that theload currents suppliediby saidamplifier to said-load arein phase.
30. In an .amplifyingsystem,a source ofcamplitude modulated carrieroscillations, a first amplifier having a grid, a cathode and anode,-.a load circuit, a circuitcompTi'sing -reactance elements so arranged and; proportioned =that' the phase constant of saidcircuit is anroddrintegral multiple of 52 radians andhaving' input terminalscconnected to the anode-cathode circuit of said amplifier and output terminals connected to said load :cir-
cuit, a secondamplifier having a: grid, a cathode andan anode, a circuit-connecting-the.-.anodecathode circuit-ofsaid second amplifier tosaid load circuit and having a phase constant of an even integral multiple-of the carrier level the outputof saidsecond amplifier is substantially linearly proportionalto said voltages and the current supplied to said load circuit bysaid amplifiers are in phase.
31. In combination, a loadzcircuit, and an amplifying .system' for supplying modulated waves thereto comprising a vacuum tube having an input circuit and an output circuit, a quarter wave network connecting said output circuit to s'aid load circuit, a second vacuiim .tube, and means for so connecting said secondvacuum tube" to said load circuit that for outputs above a pree determined value only, current will be supplied thereby to said load circuit in phase with the current supplied by said first vacuum tube.
32. An amplifying system for supplying signal modulated carrier waves comprising two space discharge devices each having. an input circuit and an output circuit, an outputnetworkcomprising Y a substantially non-dissipative, quarter wave-length passive transducerand a work .circuit coupled to said transducer'at lone 'endithereof, circuits coupling the -output circuitsoof said discharge devices to :theopposite ends of said network respectively, andwave source. means 'in the input circuits of said-devices :"impressing oscillatory voltages thereon, A the voltage in-(said input circuits being j so related-as to produce by their conjoint action highdrequency oscillations varying in amplitude in I accordance with the signal at the output terminals #of one-of saiddeviceS and at 2' the terminals of said work x-circui't, and throughoutyat least a. portion ofthe signal cycle, a substantially constant high frequency voltage at the terminals of the other of said devices. s
33. An amplifying system forsupplying high frequencysignal modulated carrier waves comprising two vacuum tubes each ,havingan anode, a cathode and a control grid, an output network comprising a substantially non-dissipative quarter wave-length passive transducer and a work circuit constituting a load impedance coupled to said transducer at one end thereof, circuits coupling theanode space paths of said tubes to the opposite ends of said network respectively, input circuits connected respectively to the grids of said tubes, and wave source means included in saidinput circuit for impressing oscillatory voltages on said grids'to produce in said work circuit high frequency oscillations modulated in amplitude in accordance withv a signal, the voltages impressed on said grids being so: related as to produceby their conjoint. action -asubstantially constant high frequency voltage across the anode space path of one of saidttubes in the presence of varying currents in the anode. space path of the other of said tubes.
34. An amplifying system -for:supplying high frequency waves modulated in amplitude in accordancewith a signal comprisingtwo space dischargedevices, each-having an anode, a cathode and a control grid, an output network comprising a substantially non-dissipative quarter wavelength passive transducer and a work circuit constituting a'load impedancecoupled to said transducer at woneend thereof, circuits coupling the anode space paths of saiddischarge-devicesto the" "opposite :ends of said network respectively, input circuits connected respectively to the grids of said discharge devices, andwavesource means included in" said input circuits impressing oscil latory voltageson saidgrids whereby-highsfrequencycurrent. pulsations of amplitude varying with the signals .arexproduced in :the anode space paths of said dischargedevices,.the voltages impressed on said grids producing by their conjoint action, voltages of signal modulated: amplitudes at .theHoutput-terminals of one of said devices and atthe terminals of saidwork ,circ-uit and'throughout atleast a portion of the signalcycle; a substantially constant output. voltage at the output terminals of the otherofsaid discharge devices. 1
=35. .In anzamplifying system forasupplying high frequency'wavesimodulatedin amplitude in .accordance with asignal, a 1 space Y discharge .1 device having input andoutput. circuits, means-fincludi-ng a high'frequency wave source in-the input circuit of said 1 device for producing I in rthe output-circuit thereof carrier frequency current impulses varying in amplitude in accordance with the signal, asecond space sdischargedevice having input and output circuits, means in the input circuit ofsaid' seconddevice forimpressing- -=oscillatory' voltages thereon. and producing in the output circuitl thereof high frequency cur rent impulses in phase quadraturewith the cur rent impulseseof said: first 'device; a- Work circuit, and circuits'coupling the output -.ci-r'cuits of-said devices to said. work circuit, one of said; coupling circuits including a quarterwave-zlength reactive network whereby the output current ;-:of said'cdethe voltages impressed on' said space: :discharge devices being so relatediin their-magnitudes and variations as to produce by their conjoint action voltages of signal modulated amplitudes at the output of one of said devices and at the terminals of said work circuit and, at least during a portion of the signal cycle, a substantially constant volt-age at the output of the other of .said devices.
36. An amplifying system comprising in com.- bination a load circuit, a space discharge device having an input circuit and an output circuit, a quarter wave-length'network coupling said output circuit to said load circuit, a second space discharge device having an input circuit and an output circuit, a circuit coupling the output circuit of said seconddevice to said load circuit, and means impressing oscillatory voltages on the input circuits of said devices whereby high frequency currents of signal dependent amplitudes are produced in theoutput circuits thereof, the voltages impressed on said input circuits being so related that the currents from the said output circuits combine in phase with each other in said load circuit and that the high frequency current in the output of one of said devices is zero in the absence of signal variations.
37. An amplifying system in accordance with claim 36 in which the voltages in said input circuits are so related as to produce a voltage of signal modulated amplitude at the output terminals of one of said devices, and, during at least a portion of the signal cycle, a voltage of substantially constant amplitude at the output terminals of the other of said devices.
38. A system for supplying amplitude modulated waves comprising a plurality of vacuum tubes, a load circuit, means causing one of said tubes to operate with a variable voltage at its output terminals, and means cooperating'with said one tube for causing another of said tubes to operate during at least a portion of the modulation cycle at a substantially constant output voltage while supplying varying current energy. 39. A system for supplying amplitude modulated waves comprising a plurality of vacuum tubes, a load circuit, means causing one of said tubes to operate with an oscillating voltage of variable amplitude at its output terminals, and means including a quarter Wave-length network cooperating with said one tube for causing another of said tubes to operate during at least a portion of the modulation cycle at a substantially constant oscillatory output voltage while supplying varying current oscillatory energy.
40. The method of operating a multiple tube amplifier at high efficiency for supplying amplitude modulated oscillations to a load circuit which comprises supplying the oscillations from the plurality of Vacuum tubes, operating one of said tubes to supply varying energy at a substantially constant output voltage during at least a portion of the modulation cycle, and operating a second tube at a varying output voltage.
41. The method of operating at high efficiency an amplifier comprising a plurality of vacuum tubes for supplying signal modulated oscillations to a load, which comprises operating one vacuum tube at a constant oscillatory output voltage and varying oscillatory output current during at least a portion of the modulation cycle, converting said variable current-constant voltage oscillations into constant current-variable voltage oscillations, supplying such converted oscillations to the load, operating a second vacuum tube at an oscillatory output voltage of varying amplitude, and combining the outputs of said tubes.
42. In-an amplifying system for supplying amplitude modulated carrier waves, an output network comprising a quarter wave-length transducer and a work circuit connected at one end of said transducer, space discharge means supplying constant amplitude high frequency oscillations to one end of said output network during at least a portion of the modulation cycle, and other space discharge means supplying to the other end of said network high frequency oscillations of signal dependent ampitudes and in quadrature phase relation to the oscillations supplied to said network from the first discharge means.
43. In an amplifying system, a work circuit, space discharge means coupled to said work circuit and supplying thereto high frequency oscillations of constant amplitude during at least a portion of the cycle of operation, a second space discharge device separately coupled to said Work circuit and supplying thereto high frequency oscillations of signal dependent amplitudes, a quarter wave-length network included in the coupling between one of said devices and said work circuit and means controlling the relative phases of the oscillations from said devices whereby they combine in phase in said work circuit.
44. In an amplifier for supplying to a load circuit high frequency oscillatory energy of amplitudes varying in accordance with a signal, a vacuum tube having an output circuit, a passive transducer having two sets of terminals and of such characteristics that the impedance measured at one set of terminals is the reciprocal of the impedance of a circuit connected to the other set of terminals, a circuit for supplying power from said output circuit to the load circuit and including connections from said output circuit to one of said sets of terminals, a second vacuum tube having an output circuitconnected to the other of said sets of terminals, and means for so exciting said second vacuum tube that the impedance-of its output circuit varies in accordance with the-signal at least throughout a portion of the signal cycle, and the current in said output circuit is in phase quadrature with the current in the output circuit of the first vacuum tube.
45. An amplifying system for supplying high frequency waves modulated in amplitude in accordance with a signal, comprising two space dis charge devices each having an anode, a cathode and a control grid, a substantially non-dissipative quarter wave-length passive transducer having one terminal connected to the cathode-anode circuit of one of said devices, a work circuit constituting a load impedance connected to the other terminal of said transducer in parallel with the anode-cathode circuit of the other of said devices, input circuits connected respectively to the grids of said devices and wave source means included in said input circuit impressing oscillatory voltages on said grids whereby high frequency current pulsations of amplitudes varying with the signal are produced in the anode space paths of said devices, the voltages impressed on said grids producing by their conjoint action voltages of signal modulated amplitudes at the terminals of said work circuit and during at least a portion of the signal cycle, a substantially constant output voltage across the anode space path of said one of said devices.
46. An amplifying system for supplying high frequency waves modulated in amplitude in aced to the anode-cathode circuit. of one of said discharge devices and a second set of terminals connected to said work circuit and the anodecathode circuit of the other of said devices in parallel, input circuits connected respectively to the grids of said discharge devices, and wave source means included in said input circuits for impressing oscillatory voltages on said gridsv whereby high frequency current pulsations of amplitudes varying with the signal are produced in the anode space paths of said discharge devices, the voltages impressed on said grids producing by their conjoint action a substantially constant voltage across the anode space path of 20 said one of said discharge devices in the presence of varying current in the anode space path of said other of said discharge devices.
47. An amplifying system comprising two space discharge devices having output circuits and input circuits, 2. load circuit; circuits respectively coupling the output circuits of said amplifiers to said load circuit, impedance inverting network means included in one of said coupling circuits,
wave source means impressing oscillations on the input circuits of said space discharge devices to produce in the output circuits thereof and in said load circuit high frequency currents varying in amplitude in accordance with a signal, and means controlling the relative amplitudes of the impressed oscillations whereby the output Voltage of one of said discharge devices is maintained substantially constant in the presence of currents of ,varying amplitude in the output circuit of the'other of said devices.
WILLIAM H. DOI-IERTY.