US2238236A - Modulation system - Google Patents
Modulation system Download PDFInfo
- Publication number
- US2238236A US2238236A US278526A US27852639A US2238236A US 2238236 A US2238236 A US 2238236A US 278526 A US278526 A US 278526A US 27852639 A US27852639 A US 27852639A US 2238236 A US2238236 A US 2238236A
- Authority
- US
- United States
- Prior art keywords
- tube
- grid
- power
- voltage
- modulation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03F—AMPLIFIERS
- H03F1/00—Details of amplifiers with only discharge tubes, only semiconductor devices or only unspecified devices as amplifying elements
- H03F1/02—Modifications of amplifiers to raise the efficiency, e.g. gliding Class A stages, use of an auxiliary oscillation
- H03F1/04—Modifications of amplifiers to raise the efficiency, e.g. gliding Class A stages, use of an auxiliary oscillation in discharge-tube amplifiers
- H03F1/06—Modifications of amplifiers to raise the efficiency, e.g. gliding Class A stages, use of an auxiliary oscillation in discharge-tube amplifiers to raise the efficiency of amplifying modulated radio frequency waves; to raise the efficiency of amplifiers acting also as modulators
- H03F1/07—Doherty-type amplifiers
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03F—AMPLIFIERS
- H03F1/00—Details of amplifiers with only discharge tubes, only semiconductor devices or only unspecified devices as amplifying elements
- H03F1/02—Modifications of amplifiers to raise the efficiency, e.g. gliding Class A stages, use of an auxiliary oscillation
- H03F1/04—Modifications of amplifiers to raise the efficiency, e.g. gliding Class A stages, use of an auxiliary oscillation in discharge-tube amplifiers
- H03F1/06—Modifications of amplifiers to raise the efficiency, e.g. gliding Class A stages, use of an auxiliary oscillation in discharge-tube amplifiers to raise the efficiency of amplifying modulated radio frequency waves; to raise the efficiency of amplifiers acting also as modulators
Definitions
- Another arrangement for delivering modulated carrier-wave power with comparatively high power efliciency is based upon the principle of plate modulation of the output stage.
- the additional power handling requirement of the modulating frequency driver tube in such a system reduces the amount oi output power which can be obtained from a given number of output tubes of a given power handling capacity, or in other words, reduces the socalled tube efficiency of the system.
- both thepower efficiency and the tube efficiency of a system for delivering modulated carrier-wave power are increased as compared with either of the above outlined systems heretofore known.
- the system of the present invention presents still greater advantages in respect to power efficiency-and tube efficiency.
- my invention comprises a grid modulated output stage having two separate tubes or sets of tubes coupled through an impedance inverting network and biased so that only the rst of the two tubes is operative at output levels below the normal carrier level, but so that when the output level rises above the normal carrier level the second tube commences to operate and simultaneously serve to vary the ⁇ effective load impedance offered.. to the rst tube.
- both tubes are operated as class C amplifiers2 thus making possible theoretically an approach to the ideal class C power eciency of 100%.
- FIG. 1 is a simplied circuit diagram illustrating the basic principles of my invention
- Fig. 2 is a more detailed circuit diagram representing a preferred embodiment of my invention.
- Fig. 3 is a set of curves illustrating the operation of an embodiment employing the circuit of Fig. 2.
- I and 2 represent the two tubes or two sets of tubes which, in accordance with my invention, are used in the output stage of the system.
- These two tubes I and 2 are supplied with direct current plate voltage from a source (not shown) in any conventional manner.
- the grids of both these tubes Vare biased far below the cutoff voltage for the corresponding plate supply, so that in the absence of signal applied to these grids neither tube passes current.
- the grid bias voltage of tube I is -Ebl and the corresponding grid bias voltage of tube 2 is -Eb2.
- Signal voltages -I-ESI and -l-EsZ of modulating frequency, such as audio-frequency signals for example, are supplied preferably cophasally to the grids of tubes I and 2 as shown.
- bias potential -EbI should exceed the peak amplitude of the audio-frequency signal voltage -EsI sufficiently so that even at the most positive peaks of this audio-frequency voltage the grid of tube I is substantially more than sumciently' negative to substantially block the plate current.
- bias potential E292 should exceed the peak amplitude of the audiofrequency voltage E32 sufliciently so that at the most positive peaks of this audio-frequency voltage the grid of tube 2 does not rise above cutoff.
- unmodulated carrier frequency potentials Erl and Er2 are applied in phase quadrature to the two grids in such sense as to compensate for the 90 degree phase shift of network 3, so that the outputs of the two tubes will be in phase at load 4.
- the various grid voltages Ebl, Eb2, ESI, E32, Erl, vE12 are related in a particular manner so that both tubes I and 2 operate as class C ampliers at all times.
- the relationships required to attain this result are somewhat complex, especially for tube I, owing to the fact that the effective load across tube I begins abruptly to fall, as hereafter explained, when the output exceeds normal carrier level. This causes the radio frequency Voltage on the plate of tube I to follow an irregular course.
- the particular manner in which class C operation is attained in spite of these diiculties can best be described later on Ll O after the general operating conditions have been considered.
- the plate circuits of the tubes I and 2 are joined together over an impedance inverting network 3, which may be considered as a transmission line having a surge impedance R, and an electrical length of one quarter wavelength at the carrier frequency, but which in practise is preferably a network of lumped impedances C, L and C as shown, at least for the medium wavelengths ordinarily used in present day broadcasting.
- the load 4 may be an antenna or any other useful load, but has been represented as a lumped element having a purely resistive impedance of the value R/Z. In the system of Fig. l thisv load is connected directly across the second tube 2.
- one of the two tubes is connected directly across the input terminal of the impedance inverting network 3, whereas the load and the other of the two tubes are connected either in series or shunt with one another across the output terminal of such network 3.
- the tube which is connected directly across the input terminals of the network is tube I, i. e. the normal or constantly operating tube, while the load 4 and the occasionally operating or secondary tube 2 are connected in shunt across the output terminals of the network.
- This relationship of the four principal elements corresponds to the relationship suggested by W. H. Doherty in the article above referred to. Other relationships, however, are possible.
- the secondary tube 2 may be connected acrossthe input terminals of the network 3 and then the rst tube and load may be connected in series with one another across the output terminal of said network.
- any of the relationships suitable for use in the system proposed by W. H. Doherty in the article above referred to may be used in the system of the present invention.
- the necessary and suicient condition is merely that the first or constantly operating tube should be so related to the load 4 and to the additional elements 2 and 3, that when the secondary tube 2 comes into operation and begins to deliver power to the load the apparent impedance presented to the output of the first tube will fall so as to enable the first tube to operate with rising current but substantially constant voltage during the intervals of rising power when the output level is being increased upward from the normal carrier value.
- a variable ⁇ load impedance is provided whi-ch increases as the modulation cycle goes from peak to carrier level, so that the alternating voltage between the plate and the cathode remains constant and the Veiliciency high.
- Such a non-linear impedance is supplied bythe combination of quarter-wave line 3 and .tube 2 shown in Fig.q 1.
- a quarter-wave line neglecting losses, has the property of transforming power at constant voltage and variable current into power at constant current and variable voltage. Since it is desired to have the voltage delivered to the load 4 vary while, in tube I, the voltage remains ⁇ constant and the current varies, the quarter-wave line is a useful connecting link between the two.
- Tube 2 which is biased so that it does not begin to pass -plate current until normal carrier level is reached, serves the double purpose of furnishing a variable impedance for tube I and supplying additional power during the positive peaks of the modulation cycle.
- the first step n applying impedance inversion to grid modulation is to calculate or otherwise determine the proper load impedance R which tube I should work into as a class C amplier at peak level in order to deliver maximum possible power output consistent with allowable plate dissipation, available cathode emission, and allowable direct plate voltage.
- This value of plate dissipation can be made higher than for unmodulated class C operation because the tube is operating at peak level only :a small fraction of the time.
- the allowable direct plate voltage can be made higher than for a plate-modulated class C amplifier because the instantaneous plate-supply voltage does not rise during modulation and put additional voltage strains on tube and apparatus, as it does in a plate-modulated amplifier.
- the actual load, 4, placed in parallel Wit-h tube 2 is made equal to R/2.
- the characteristic impedance of the quarter-wave line is made equal to R; that is, the reactance of inductance L at the carrier frequency is R, and the reactance of each of the condensers C is also equal to R..
- Signal, (i, e., modulating) voltages ESI and E52 are applied to the two tubes in the same phase.
- the load impedance on tube I at carrier level conditions is equal to the square of the characteristic impedance divided by the actual load fimpedance, or 2R.
- This ⁇ highload impedance enables tube I to work at twice the efficiency which' it otherwise would have Afor normal carrier level or lower'levels. It will be noted that this causes the efciency at normal carrier level to be as great as in a normal unmodulated or plate-modulated class C amplifier, instead of only half this great as in the case of an ordinary grid-modulated amplifier.
- tube 2 starts to pass plate current in the form of short pulses, which increase in magnitude until, at peak level, this tube is also delivering its full rated output to theload 4. Since tube 2 delivers power to the load, it presents an apparent negative resistance across the output of the impedance inverter in parallel with the actual load. This causes the apparent load resistance presented to the output terminals of t-he impedance inverter 3 to increase as the modulation cycle goes from carrier to peak level, which means that the input resistance of the inverter decreases. Tube I is thus able to supply a constantly increasing amount of power without increasing the radio-frequency plate voltage (which was already at its maximum possible value) as the modulation goes from carrier to peak level.
- tube I operates as a class C famplier with a relatively large alternating voltage between the plate and the cathode most of the time, and so has high average plate effi-A where Et is the peak value of the radio frequency voltage developed across the load 4 at maximum output (i. e.
- al and ,a2 are the amplication factors ofthe respective tubes;
- Eco is the negative grid voltage for cutoi ⁇ with undiminished supply potential on the anode;
- -E'cion is the grid cutoff when the anode potential is diminished by of Et;
- -E'cto isthe grid cutoff when the anode potential is diminished by 50% of Et;
- -Ecx, -Ecy and -Ecz are the corresponding cutoffs when the anode potential is diminished by (l-X) Et,
- the factors (1-X), (1-Y) and (l-Z) are decimal fractions whose arc cosines equal half the duration of the pulses of plate current for tube 2 at peak audio signals, for tube I at Zero audio signal (i. e. normal carrier level), and for tube I at peak audio signal respectively.
- (1-X) were .819 then during peak audio signals the plate pulses of tube 2 would begin and end at those instants when the radio frequency potential-Em ⁇ on the grid (and likewise the radio frequency swing of the load cir cuit) had 81.9% of their respective peak values. This would mean that each pulse would begin 35 before the radio frequency peak and would end 35 after the radio frequency peak, thus lasting '70 or 7/31; of a cycle.
- (1-Y) Were .94 and (l-Z) were .766, ⁇ then the plate pulses of tube I would last 1A) of a cycle when the audio signal was zero, and of a cycle when the audio signal was a maximum.
- the load voltage Et may b-e found by subtracting from Ep the drop required to produce the desired plate current without necessitating an unreasonably positive grid. Or with ordinary triodes Et may be simply assumed to be 80% of Ep. A load resistance value 2R is then chosen which, at the assumed radio frequency Voltage Et, will draw the normal amount of power desired for mean carrier level (i. e. one-quarter the desired peak output power).
- the tube I should be capable of delivering twice this normal amount of power, at least intermittently without overheating. 'Ihe maximum current density attainable with this tube should also be suiicient to produce the desired normal power even with a slightly subnormal pulse length.
- Equation l0 The selection of Y and Z is governed by the same limitations as the selection of X and in addition is further restricted by the fact that Y and Z are interdependent as shown in Equation l0. As is apparent from Equations 7-10, Z must exceed 2Y; and in fact it can be seen that it is advantageous to make this excess very considerable so far as consistent with good efficiency and power output, since the several p0- tentials Erl, ESI, EbI, PI vary more or less inversely With Z2Y, and since it is clearly advantageous to keep these potentials reasonably low.
- arc cosine (l-X) roughly equal to arc cosine (l-Z) if tubes I and 2 are alike (or to proportion these angles inversely as the maximum current densities of the tubes if the tubes are different) in order to equalize the tube outputs at peak audio signal level.
- PI is determined by the choice of Y and Z and is likely to turn out higher than desired unless Y and Z are carefully chosen. By increasing the difference between Y and Z, however, PI may be reduced, and a compromise set of values can thus be arrived at which will give good eiiiciency and output Without an impracticably high positive grid limit PI. In fact a substantial range of values of XYZ and PI will be found which will give satisfactory operation. As for P2 there is a still greater latitude.
- the regulation of the source of Erl is deliberately made poor so that the voltage Erl will fall off considerably during the flow of grid current.
- This feature reduces the variations of load on the source of ESI and furthermore is advantageous where coated or thoriated cathodes are employed, or where for any other reason a high positive value of PI is objectionable.
- the poor regulation may be effected in any well known manner, for example, by designing the source of voltage Erl to have high impedance, or by feeding this voltage to the grid over a high impedance or better yet a quarter wave line of high surge impedance.
- the element causing the poor regulation should not be common to ETI and E12 nor to EN and EsI, since it should preferably not influence anything but ETI.
- the Formulae l-13 are still applicable though.
- PI Will no longer represent the actual peak grid voltage (with respect to Eco) but .Will only measure the sharpness of transition at the start and end of the pulses.
- a grid leak resistor and condenser are used to supply at least part of the bias EbI for tube I.
- This is advantageous in a number of ways; for instance, it generally results in a considerable reduction in the required amplitude of Erl, a relaxation of the requirement that Z should considerably exceed 2Y, a marked reduction in PI, an increase in latitude in the selection of all the voltages for the grid of tube I, and usually an increase in emciency or power output as compared with the system above described in conjunction with Fig. 1.
- Equations '7-10 are replaced by the following:
- wnere Eli. represents the potential drop across the grid leak during a maximum of the audio signal, and A represents the corresponding drop when the audio signal has zero value. It will benoted that now a very wide latitude is possible in selecting Y, Z and PI since it is now only necessary that Z exceed Y (instead of that it exceed ZY).
- Y and Z should be kept smallsay below .'75 and preferably below .55.
- Yy and Z should be large enough so that the corresponding pulse lengths permit full 'power output without excessive current densities.
- the pulse lengths for X, Yand Z should each exceed 60.
- FIG. 2 A preferred circuitfembodying my invention is shown in Fig. 2.
- the 90 phase shift producedby the quarter-wave line 3 in the plate circuit is compensated for by an equal and opposite 'phase shift produced by a quarter-wave line 5 of opposite type in the grid circuit of tube I.
- the quarter-wave lines 3 and 5 have for their shunt reactances, parallel resonant circuits CI-LI, C2--L2, C4--L4 and C ⁇ 5-L5 detuned to give the required reactances.
- this circuit incorporates both the poor ⁇ - regulation feature land the'part-ia'l grid leak bias feature above discussed.
- Leak-resistor RI provides part of the bias voltage 4for tube I.
- the impedance of grid line 5y with its terminating resistance R2 isy sufficiently-highA to cause voltage Erl to be greatly ⁇ reduced by grid current. 1
- the modulating voltage is applied to thegrids of tubes I and 2 in the same phase, but with different magnitudes, by means of the tapped transformer T.
- triodesEsZ may be betweenone-half and two-thirds of Esl.
- the voltages Erl and E ⁇ r2' may. belequal ⁇ or un-z equal according to the adjustment of the resistance R2 terminating the grid phasing line 5.
- This resistance may or may not be required,depending on how much it is desired to let the grids go positive, and the ⁇ characteristic impedance of ⁇ the grid line 5.
- CB is the series element of the grid line 5 and its reactance is made equal to the desired characteristic impedance of this line.
- L3 is the series element of l the impedance inverting line 3 and corresponds to the inductance L in Fig. l.vv CnI and -C712, are theneutralizing condensers.
- .AI-A4 are direct ⁇ current meters for plate. and grid currents
- BI-B are blocking condensers having negligible impedance atl carrier frequency butserving to block direct currents, and in the case of condensers BI, B2,.B3 and B5 also serving tocblock audio frequencies.
- the grid line 5 should be adjusted to have a phase shift of exactly degrees.
- a cathode-ray oscillograph could be used by ⁇ connecting onefpair of plates across each vend of ,the network uand adjusting. for acircle on the screen, but the practice of this method requires a certain amount of skill because of the presence of harmonics in the voltages. Also the capacitance of the oscillograph disturbs conditions unless it is to beleft permanently connected in the circuit.
- a better method, which does not require the connection of any apparatus to the radio-frequency circuits, is to couple the radio-frequency input terminals TI, T2 loosely VIto the driver stage DS, after having xed C6 at the desired value as explained above, 4and then ⁇ vary Cll while watching the plate-current meter (not shown) of the driver tube.
- a maximum will be passed through at series yresonance between C6Y and 1.404, which is the ⁇ correct adjustment.
- ⁇ Before the final adjustment of C4 ⁇ is made the neutralizing condensers Cul and (In-2 ⁇ should be adjusted, which ⁇ can be done by any-of the -conventional methods.
- CELS should also be adjusted to resonance, but is .not critical as it does not affect the relative phases of the grid voltages.
- ⁇ i f V The radio-frequency excitation voltages on the grids of the two tubes I and 2 are preferably adjusted to the calculated values without connecting any measuring device, such as a vacuum-tube voltmeter, which might have 'suflici'ent capacitance to change the tuning and disturb the voltages -being measured. This can be ac ⁇ complished by using the'power tubes I and 2 as vacuum-tube voltmeters. To do this the direct-voltage supply to the plates of I and 2 is disconnected, the radio-frequency voltage is applied to the input terminals TI and T2, and the negative grid bias on each tube increased until grid current just ceases to ow. 'Ihe peak radiofrequency voltages are then equal to the bias potentials.
- the rst is that impedances must be accurately inverted (i. e., the line must be a quarter-wave line), while the second condition is that the characteristic impedance must also be correct.
- the characteristic impedance is determined by the reactance of L3.
- the equivalent series reactance of L3 will not be the value computed from the inductance, but will be somewhat greater because of partial resonance with the distributed capacitance of the coil and wiring.
- Some method is hence needed for experimentally determining the proper value of L3 under actual conditions.
- a cathode-ray tube could theoretically be used, but has the disadvantage of introducing capacitance unless it be left permanently in circuit. Small harmonic voltages across the tank circuits also cause trouble when a cathode-ray tube is used.
- a simpler and more satisfactory technique is as follows:
- a load resistance of the proper Value is rst connected directly across the output L2C2. If the value previously computed as correct for each tube at peak level is denoted by R, the amount of resistance used is R/2.
- the lament of tube 2 is then opened, but this tube is left in the socket so that its capacitance will be present. Since the tube has been neutralized, the tube capacitances will be the same with the filament cold as when it is operating.
- the normal value of radio-frequency grid excitation is applied, and the direct grid voltage of tube I set at any convenient Value. Plate voltage is applied, preferably of reduced value or through a protective resistance. Condenser CI is then adjusted for minimum plate current of tube I as in any class C amplier, while C2 is adjusted for maximum plate current.
- This second minimum will, in general, be at a dierent capacitance on CI and will be a diierent value of direct plate current. may be changed in steps of one or two divisions while varying CI slowly back and forth until the particular combination giving the highest possible minimum is found, which is the correct combination to use. In carrying out this procedure the minimum for each setting of C2 should be obtained and recorded. It is not desir- CII able to adjust CI and C2 alternately to get the highest minimum as the adjustments then diverge away from the correct value.
- CI can be changed in steps and C2 continuously varied, when a. series of maxima will be observed for the plate current of tube I.
- the correct tuning in this case is the value of CI which gives the lowest possible maximum as C2 is Varied. Both ways will arrive at the same Values of capacitance.
- the above tuning procedure insures that the impedance inverter 3 is working properly, but it does not verify the correctness of its characteristic impedance.
- the characteristic impedance can, however, be determined by shifting the load from one end of the inverter to the other as follows: With the filament of tube 2 opened and that of tube I closed, and a load of R/2 across tube 2, the condensers CI and C2 are tuned as explained above and the plate current of tube I is noted. Then the series element L3 is opened and a resistance 2R is placed directly across LICI, which is retuned for minimum plate currentin tube I. The minimum plate current should then have the same Value as before. If it does not the inductance of L3 was wrong. L3 may then be changed slightly, and the process repeated until R/2 across the output of the inverter draws the same plate current as 2R across the tube directly.
- the grid leak RI may be changed, or the ratio of radio-frequency grid voltages can be changed slightly by adjusting R2. Then, as modulating voltage is applied, the radio-frequency load current should change linearly. Finally R/2 is replaced by the antenna or other desired load by coupling the antenna to the tank circuit L2C2.
- the platel currents of tubes I and 2 are respectively plotted against the instantaneous modulating voltage ESI.
- Curve d shows that even with small receiver type tubes and correspondingly low plate potentials and powers the momentary eciency is above 60% not only in the unmodulated condition but also throughout the modulation cycle provided the amplitude modulating signal Ecl does not exceed volts (corresponding to about 18% modulation as seen from curve c) Furthermore, even during periods of 50% modulation it will be seen that the momentary efficiency remains above 60% during about two-thirds or three-fifths of eachl audio cycle and remains above 50% during almost the whole of each cycle. The average eiciency even at 50% modulation is therefore well above 50%.
- the average efficiency is again higher because the power handled during modulation peaks so far outweighs the power handled during modulation troughs that the relatively high e'ciency at the peaks becomes the controlling factor which principally determines the average efficiency at high percentages of modulation.
- Curve c is evidently a substantially straight line, thus showing clearly the practical linearity of the modulation envelope with respect to the modulating audio signal
- the radio frequency and audio frequency signals applied to the grids may be a single cornplex voltage instead of two independent voltages.
- the predetermined signal level at which it begins to operate is equal to zero audio signal and corresponds to normal carrier level (i. e. one-half the carrier amplitude of peak output) it is possible to displace this level so that tube 2 comes into operation at a somewhat higher or lower level.
- the powers and/ or the anode supply voltages for the two tubes may differ.
- unequal power division and/or supply voltages may be employed even when tube 2 comes into operation at the half-way point of the audio signal as described.
- my invention contemplates grid modulation and class C operation of tube I in a circuit of the type wherein tube 2 is inoperative at low levels and wherein by virtue of impedance inversion the effective load on tube I is sharply reduced when this tube 2 does come into action.
- grid modulation could be applied to such a tube as tube I, whose effective load is subject to discontinuous variation
- class C operation was possible with such a tube nor how such class C operation could be attained.
- the fact that both grid modulation and class C operation could be used in such a system and the further fact that such a system is inherently roughly linear and can readily be made accurately linear are, so far as I am aware, totally unexpected.
- a system for delivering modulated carrierwave power to a load which comprises a iirst and a second discharge tube each having a cathode, a grid and an anode, means for supplying continuous potential to said anodes, connections for 4 delivering power from the anodes of both of said tubes to said load, an impedance inverting network connected between the anode of one of said of said first discharge tube an impedance whose value decreases with increasing power output' from said second discharge tube, means for applying carrier frequency signals to the grids of both said tubes, means for applying modulation frequency signals to the grids of both said tubes, means for biasing said second tube to be inoperative at all values of said modulation signal below a predetermined value and to operate in class C at all values of said modulation signal above said predetermined value, and means for biasing said rst tube for class C operation at all values of said modulation signal.
- a system for delivering modulatedcarrierwave power to a load which'comprises a iirst and a'second discharge tube each having a cathode, a grid and an anode, means for supplying continuous potential to said anodes, connections for delivering power from the anodes of both of said tubes to said load, an impedance inverting network connected between the anod-e of one of said tubes and said load so as to present to the anode of said first discharge tube an impedance whose value decreases with increasing power output from said second discharge tube, means for applying carrier and modulation frequency signals to the grids of both said tubes, and means for establishing a substantially linear characteristic, comprising means for applying to the grid of said first tube a bias potential sufficient tovjust cut ofi the flow of current at the most negative value of said modulation signals and to cause said first tube to pass current in short pulses of less than t6 of a cycle duration during intervals of zero value of said modulation signals, and means for applying to the grid of said second
- a system for delivering modulated carrierwave power to a load which comprises a first and a second discharge tube each having a cathode, a grid and an anode, means for supplying continuous potential to said anodes, connections for delivering power from the anodes of both of said tubes to said load, an impedance inverting network connected between the anode of said first tube and said load so as .to present to the anode of said first discharge tube an impedance whose value decreases with increasing power output from said second discharge tube, means for applying carrier and modulation frequency signals to the grids of both said tubes, and means for establishing a substantially linear characteristic, comprising means for applying to the grid of said first tube a bias potential sufficient to just cut off the ow of current at the most negative value of said modulation signals and to cause said first tube to pass current in pulses whose duration during intervals of maximum value of said modulation signals is between 1.1 and 1.5 times their corresponding duration during intervals of zero value of said modulation signals, and means for applying to the grid of said
- a system for delivering modulated carrierwave power to a load comprising a condenser and a grid leak resistor connected to produce a bias Voltage in response to the flow of grid current in said rst tube.
- a system for delivering modulated carrierwave power to a load includes a source of carrier frequency signals hav- V ing an impedance sulciently high to cause the voltage of said carrier frequency signals applied to the grid of said first tube to fall oi substantially in magnitude responsive to the flow of grid current in said tube.
- a system for delivering modulated carrierwave power wherein said means for applying carrier frequency signals to the grids of both said tubes include a source of carrier frequency signals, a phase shifting network, connections from said source of carrier wave signals directly to the grid of said second tube, and connections from said source of carrier frequency signals through said phase shifting network to the grid of said first tube, the effective impedance of said phase shifting network and source as seen from the grid of said second tube being sufficiently high to cause substantial falling off of the voltage of said carrier frequency signals on the grid of said rst tube in response to the ow of grid current in said rst tube, and wherein said means for biasing said first tube include a grid leak resistor and a condenser for providing a bias potential responsive to the flow of grid current.
- a system for delivering modulated carrier- Wave power to a load wherein the carrier frequency, the modulation frequency and bias voltages applied to ythe grid of said first tube are equal respectively to 1 Et -ZPl "t'i Y EL ZPl l-- -l- A and wherein a grid leak is also connected to the grid of the rst tube to provide a further grid bias voltage whose value during intervals of maximum power output is equal to where al is the amplification factor of the rst tube, A is the value of said further grid bias voltage provided by the grid leak during intervals of normal output, P1 has a value of the order of 1/5 of the anode supply voltage divided by 1, Et has a value of the order of 4K5 the anode supply voltage, Y has a value between .25 and .75, and Z has a value between .40 and .90.
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Amplifiers (AREA)
Description
April l5, 1941. F. E. TERMAN MODULATION SYSTEM Filed June 1o, 1959 ATTORNEY patented Apr. l5, 1941 Ult" ITE!) STATES MODULATION SYSTEM Frederick E. Terman, Stanford University, Calif., assigner to International Standard Electric Corporation, New York, N. Y., a corporation of Delaware PATENT OFFICE Application June 10, 1939, Serial No. 278,526
`8 Claims.
power available from a given amount of vacuum,-
tube apparatus is important.
It is an object of my invention to provide such a system which shall have a higher overall power efficiency than other systems of the types generally used heretofore.
It is also an object of my invention to provide such a system which shall also have an increased tube efciency, i. e. shall be capable of delivering a larger power for a given number of vacuum tubes of given size than systems heretofore known. It is a further object to provide such a system which shall be substantially free from distortion, i. e. which shall deliver a modulated carrier wave whose envelope is substantially linearly related to the modulating signal. In particular it is an object of my invention to provide a system for delivering modulated carrierwave power wherein the modulation is effected by grid modulation, thus avoiding the requirement for a considerable amount `of driving power at modulating signal frequencies.
More specifically it is an object of my invention to provide a system for delivering modulated carrier-wave power which shall approach in power efficiency, the characteristics of an unmodulated or plate modulated class C power amplifying system, and which at the same time shall require a considerably lower driving power of modulatingl signal frequency than known systems employing a plate modulated class C output stage.
Systems for delivering amplitude modulated carrier-wave power have previously been suggested in which the modulation is effected at a low power level in some stage ahead of the output stage, and in which the output stage operates as a class B amplifier or class A-B amplifier to amplify such premodulated waves. According to one suggestion heretofore advanced by W. H.
Doherty in an article entitled A New High lllfl-` ciency Power Amplifier for Modulated Waves, Proceedings I. R. E., volume 24, pages 1163,-1182, September 1936, such a class B amplifier employs two separate output tubes coupled through an impedance inverting network and excited so that at output levels below the normal carrier level only one of the two tubes is operated, whereas at levels above the normal carrier level a second tube commences to operate. This suggested system provides an efficiency considerably higher Cil than obtainable in conventional amplifiers for modulated waves but at best such a system is still subject to the inherent limitations of class B operation `and thus under at-rest conditions (i. e. with unmodulated output of normal carrier level) can theoretically only approach an ideal Yeiiiciency of 78.5% rather than the value of which is theoretically possible in unmodulated class C ampliers.
Another arrangement for delivering modulated carrier-wave power with comparatively high power efliciency is based upon the principle of plate modulation of the output stage. By virtue of `the fact that the momentary plate supply voltage in the output stage of such a system is varied in conformity with the required variation of carrier output power, it is possible to operate the output stage of such a system as a class C amplifier rather than a class B amplifier, thus rendering it theoretically possible to approach the ideal class C efciency of 100%. In such systems, however, the requirement for large amounts of modulating signal frequency power from the driver stage results in again lowering the overall power eiciency since such modulating signal frequency power cannot be efficiently generated. Furthermore, the additional power handling requirement of the modulating frequency driver tube in such a system reduces the amount oi output power which can be obtained from a given number of output tubes of a given power handling capacity, or in other words, reduces the socalled tube efficiency of the system.
In accordance with the present invention both thepower efficiency and the tube efficiency of a system for delivering modulated carrier-wave power are increased as compared with either of the above outlined systems heretofore known. As compared with conventional systems employing either low level modulation followed by ordinary amplification or grid modulation in the output stage, the system of the present invention presents still greater advantages in respect to power efficiency-and tube efficiency.
Briefly my invention comprises a grid modulated output stage having two separate tubes or sets of tubes coupled through an impedance inverting network and biased so that only the rst of the two tubes is operative at output levels below the normal carrier level, but so that when the output level rises above the normal carrier level the second tube commences to operate and simultaneously serve to vary the` effective load impedance offered.. to the rst tube. Preferably both tubes are operated as class C amplifiers2 thus making possible theoretically an approach to the ideal class C power eciency of 100%. Furthermore, because of the effectively varying load impedance on the first tube it is possible in accordance with my invention to approach reasonably closely this ideal class C power efciency, at least during those intervals when the modulating signal is small or zero so that the output level is at or near the normal carrier level, as well as during those intervals when the output power level is at or near four times the normal carrier power level (i. e. when the amplitude is twice the normal carrier amplitude), as in the case of modulation peaks at 100% modulation.
The exact nature of my invention and the mode of operation thereof may best be understood from the following description taken in conjunction with the attached drawing, in which Fig. 1 is a simplied circuit diagram illustrating the basic principles of my invention;
Fig. 2 is a more detailed circuit diagram representing a preferred embodiment of my invention; and
Fig. 3 is a set of curves illustrating the operation of an embodiment employing the circuit of Fig. 2.
Referring now more particularly to Fig. 1, I and 2 represent the two tubes or two sets of tubes which, in accordance with my invention, are used in the output stage of the system. These two tubes I and 2 are supplied with direct current plate voltage from a source (not shown) in any conventional manner. The grids of both these tubes Vare biased far below the cutoff voltage for the corresponding plate supply, so that in the absence of signal applied to these grids neither tube passes current. The grid bias voltage of tube I is -Ebl and the corresponding grid bias voltage of tube 2 is -Eb2. Signal voltages -I-ESI and -l-EsZ of modulating frequency, such as audio-frequency signals for example, are supplied preferably cophasally to the grids of tubes I and 2 as shown. The bias potential -EbI should exceed the peak amplitude of the audio-frequency signal voltage -EsI sufficiently so that even at the most positive peaks of this audio-frequency voltage the grid of tube I is substantially more than sumciently' negative to substantially block the plate current. In similar manner bias potential E292 should exceed the peak amplitude of the audiofrequency voltage E32 sufliciently so that at the most positive peaks of this audio-frequency voltage the grid of tube 2 does not rise above cutoff.
In addition to the audio-frequency signal and bias potentials above mentioned, unmodulated carrier frequency potentials Erl and Er2 are applied in phase quadrature to the two grids in such sense as to compensate for the 90 degree phase shift of network 3, so that the outputs of the two tubes will be in phase at load 4.
In accordance with an essential feature of my invention the various grid voltages Ebl, Eb2, ESI, E32, Erl, vE12 are related in a particular manner so that both tubes I and 2 operate as class C ampliers at all times. The relationships required to attain this result are somewhat complex, especially for tube I, owing to the fact that the effective load across tube I begins abruptly to fall, as hereafter explained, when the output exceeds normal carrier level. This causes the radio frequency Voltage on the plate of tube I to follow an irregular course. The particular manner in which class C operation is attained in spite of these diiculties can best be described later on Ll O after the general operating conditions have been considered.
The plate circuits of the tubes I and 2 are joined together over an impedance inverting network 3, which may be considered as a transmission line having a surge impedance R, and an electrical length of one quarter wavelength at the carrier frequency, but which in practise is preferably a network of lumped impedances C, L and C as shown, at least for the medium wavelengths ordinarily used in present day broadcasting. The load 4 may be an antenna or any other useful load, but has been represented as a lumped element having a purely resistive impedance of the value R/Z. In the system of Fig. l thisv load is connected directly across the second tube 2.
In order to more generally define the relationship of the four principal elements of my invention, it may be noted that generally one of the two tubes is connected directly across the input terminal of the impedance inverting network 3, whereas the load and the other of the two tubes are connected either in series or shunt with one another across the output terminal of such network 3. In the schematic embodiment of Fig. 1, the tube which is connected directly across the input terminals of the network is tube I, i. e. the normal or constantly operating tube, while the load 4 and the occasionally operating or secondary tube 2 are connected in shunt across the output terminals of the network. This relationship of the four principal elements corresponds to the relationship suggested by W. H. Doherty in the article above referred to. Other relationships, however, are possible. For example, the secondary tube 2 may be connected acrossthe input terminals of the network 3 and then the rst tube and load may be connected in series with one another across the output terminal of said network. In fact, any of the relationships suitable for use in the system proposed by W. H. Doherty in the article above referred to may be used in the system of the present invention. The necessary and suicient condition is merely that the first or constantly operating tube should be so related to the load 4 and to the additional elements 2 and 3, that when the secondary tube 2 comes into operation and begins to deliver power to the load the apparent impedance presented to the output of the first tube will fall so as to enable the first tube to operate with rising current but substantially constant voltage during the intervals of rising power when the output level is being increased upward from the normal carrier value.
In order to explain clearly the method of operation of the system generally illustrated in Fig. l, it will be convenient first to consider the operation of a more conventional grid modulated system such as would be formed by omitting network 3 and tube 2 from Fig. 1, and then operating tube I with the load 4 connected directly in its plate circuit. Assume that such a conventional system was operated to yield modulation, and assume that the load impedance 4 was correctly adjusted for the conditions at modulation peaks so as to give under such conditions a carrier frequency swing of plate voltage which was almost as great as the direct current plate supply voltage, thus resulting in an almost Zero value of minimum plate voltage and a correspondingly high eciency. Then whenever there was little or no audio signal or Whenever the audio-frequency signal voltage Esl was pass- 'ing through Zero, so that the output fell to normal carrier level, the plate swing would become less than half as great as the direct current plate voltage; therefore the minimum plate voltage would be more than half the direct current plate voltage, and the efficiency would be reduced to onehalf of its peak value. This low efficiency at normal carrier Ievel could be overcome by doubling the load resistance so that the minimum plate voltage would be low, and consequently the efciency high, under normal carrier level conditions. However, when the alternating plate voltage increased during modulation peaks, distortionwould result since the plate swing Awould then tend to `be greater than the direct plate voltage.
In accordance with my invention a variable `load impedance is provided whi-ch increases as the modulation cycle goes from peak to carrier level, so that the alternating voltage between the plate and the cathode remains constant and the Veiliciency high. Such a non-linear impedance is supplied bythe combination of quarter-wave line 3 and .tube 2 shown in Fig.q 1. A quarter-wave line, neglecting losses, has the property of transforming power at constant voltage and variable current into power at constant current and variable voltage. Since it is desired to have the voltage delivered to the load 4 vary while, in tube I, the voltage remains `constant and the current varies, the quarter-wave line is a useful connecting link between the two. Tube 2, which is biased so that it does not begin to pass -plate current until normal carrier level is reached, serves the double purpose of furnishing a variable impedance for tube I and supplying additional power during the positive peaks of the modulation cycle.
Before the operation of Fig. 1 can be explained in detail it` will be convenient to consider the impedance relations of the various parts of the circuit. The first step n applying impedance inversion to grid modulation is to calculate or otherwise determine the proper load impedance R which tube I should work into as a class C amplier at peak level in order to deliver maximum possible power output consistent with allowable plate dissipation, available cathode emission, and allowable direct plate voltage. This value of plate dissipation can be made higher than for unmodulated class C operation because the tube is operating at peak level only :a small fraction of the time. Also, the allowable direct plate voltage can be made higher than for a plate-modulated class C amplifier because the instantaneous plate-supply voltage does not rise during modulation and put additional voltage strains on tube and apparatus, as it does in a plate-modulated amplifier.
Having determined this value of load impedance, the actual load, 4, placed in parallel Wit-h tube 2 is made equal to R/2. The characteristic impedance of the quarter-wave line, is made equal to R; that is, the reactance of inductance L at the carrier frequency is R, and the reactance of each of the condensers C is also equal to R.. Signal, (i, e., modulating) voltages ESI and E52 are applied to the two tubes in the same phase.
The operation may now be explained as follows: Since the bias on tube 2 has such value as to make this tube inoperative between zero and carrier level, the load impedance on tube I at carrier level conditions is equal to the square of the characteristic impedance divided by the actual load fimpedance, or 2R. This `highload impedance enables tube I to work at twice the efficiency which' it otherwise would have Afor normal carrier level or lower'levels. It will be noted that this causes the efciency at normal carrier level to be as great as in a normal unmodulated or plate-modulated class C amplifier, instead of only half this great as in the case of an ordinary grid-modulated amplifier.
r `When carrier level is reached, tube 2 starts to pass plate current in the form of short pulses, which increase in magnitude until, at peak level, this tube is also delivering its full rated output to theload 4. Since tube 2 delivers power to the load, it presents an apparent negative resistance across the output of the impedance inverter in parallel with the actual load. This causes the apparent load resistance presented to the output terminals of t-he impedance inverter 3 to increase as the modulation cycle goes from carrier to peak level, which means that the input resistance of the inverter decreases. Tube I is thus able to supply a constantly increasing amount of power without increasing the radio-frequency plate voltage (which was already at its maximum possible value) as the modulation goes from carrier to peak level.
It will be noted that tube I operates as a class C famplier with a relatively large alternating voltage between the plate and the cathode most of the time, and so has high average plate effi-A where Et is the peak value of the radio frequency voltage developed across the load 4 at maximum output (i. e. during the audio frequency peak); al and ,a2 :are the amplication factors ofthe respective tubes; Eco is the negative grid voltage for cutoi` with undiminished supply potential on the anode; -E'cion is the grid cutoff when the anode potential is diminished by of Et; -E'cto isthe grid cutoff when the anode potential is diminished by 50% of Et; -Ecx, -Ecy and -Ecz are the corresponding cutoffs when the anode potential is diminished by (l-X) Et,
(1-Y) Et and(1-Z) Et, respectively.
The factors (1-X), (1-Y) and (l-Z) are decimal fractions whose arc cosines equal half the duration of the pulses of plate current for tube 2 at peak audio signals, for tube I at Zero audio signal (i. e. normal carrier level), and for tube I at peak audio signal respectively. For` example, if (l-X) were .819 then during peak audio signals the plate pulses of tube 2 would begin and end at those instants when the radio frequency potential-Em` on the grid (and likewise the radio frequency swing of the load cir cuit) had 81.9% of their respective peak values. This would mean that each pulse would begin 35 before the radio frequency peak and would end 35 after the radio frequency peak, thus lasting '70 or 7/31; of a cycle. Similarly if (1-Y) Were .94 and (l-Z) were .766,` then the plate pulses of tube I would last 1A) of a cycle when the audio signal was zero, and of a cycle when the audio signal was a maximum.
Assuming that no grid leak is employed so that 1 where PI and P2 represent the amounts by which the grids of tubes I and 2 respectively, are driven above cutoff at each radio frequency peak during a maximum of the audio signals ESI and E32.
To determine the operating conditions for given tubes and a given supply voltage Ep the load voltage Et may b-e found by subtracting from Ep the drop required to produce the desired plate current without necessitating an unreasonably positive grid. Or with ordinary triodes Et may be simply assumed to be 80% of Ep. A load resistance value 2R is then chosen which, at the assumed radio frequency Voltage Et, will draw the normal amount of power desired for mean carrier level (i. e. one-quarter the desired peak output power). The tube I should be capable of delivering twice this normal amount of power, at least intermittently without overheating. 'Ihe maximum current density attainable with this tube should also be suiicient to produce the desired normal power even with a slightly subnormal pulse length.
In selecting the values of P2 and X the same general limitations should be observed as for an unmodulated class C amplifier. If P2 is made too large, the cathode may be injured if it is of the thoriated or oxide coated type, and also the grid current may become excessive, thus heating the grid and drawing unnecessarily large amounts of power from the sources of ETI and EsI. On the other hand, if P2 is unnecessarily small the transition intervals at the start and finish of each plate current pulse (during which 7-5 intervals'current flows ineciently through the tube before or after the tube is in condition of substantially full conductivity) 1 Will be unnecessarily long. Likewise if X is large, e. g. .80, so that the plate current pulses endure for a considerable fraction of a half cycle, the efficiency will drop toward that of class B; While if X is unnecessarily small, e. g. .10, the required radio frequency driving power becomes needlessly large, and the plate current pulses will be unnecessarily brief thus reducing the power output available with a given current density.
The selection of Y and Z is governed by the same limitations as the selection of X and in addition is further restricted by the fact that Y and Z are interdependent as shown in Equation l0. As is apparent from Equations 7-10, Z must exceed 2Y; and in fact it can be seen that it is advantageous to make this excess very considerable so far as consistent with good efficiency and power output, since the several p0- tentials Erl, ESI, EbI, PI vary more or less inversely With Z2Y, and since it is clearly advantageous to keep these potentials reasonably low. Furthermore, it may be desirable also to make arc cosine (l-X) roughly equal to arc cosine (l-Z) if tubes I and 2 are alike (or to proportion these angles inversely as the maximum current densities of the tubes if the tubes are different) in order to equalize the tube outputs at peak audio signal level.
The value of PI is determined by the choice of Y and Z and is likely to turn out higher than desired unless Y and Z are carefully chosen. By increasing the difference between Y and Z, however, PI may be reduced, and a compromise set of values can thus be arrived at which will give good eiiiciency and output Without an impracticably high positive grid limit PI. In fact a substantial range of values of XYZ and PI will be found which will give satisfactory operation. As for P2 there is a still greater latitude.
In accordance with another feature of my invention the regulation of the source of Erl is deliberately made poor so that the voltage Erl will fall off considerably during the flow of grid current. This feature reduces the variations of load on the source of ESI and furthermore is advantageous where coated or thoriated cathodes are employed, or where for any other reason a high positive value of PI is objectionable. The poor regulation may be effected in any well known manner, for example, by designing the source of voltage Erl to have high impedance, or by feeding this voltage to the grid over a high impedance or better yet a quarter wave line of high surge impedance. Preferably the element causing the poor regulation should not be common to ETI and E12 nor to EN and EsI, since it should preferably not influence anything but ETI. When this feature of poor regulation is employed the Formulae l-13 are still applicable though. PI Will no longer represent the actual peak grid voltage (with respect to Eco) but .Will only measure the sharpness of transition at the start and end of the pulses.
In accordance with still another feature of my invention a grid leak resistor and condenser are used to supply at least part of the bias EbI for tube I. This is advantageous in a number of ways; for instance, it generally results in a considerable reduction in the required amplitude of Erl, a relaxation of the requirement that Z should considerably exceed 2Y, a marked reduction in PI, an increase in latitude in the selection of all the voltages for the grid of tube I, and usually an increase in emciency or power output as compared with the system above described in conjunction with Fig. 1.
When such a grid leak is used, llbll in Equations 1-6 no longer is a constant. Therefore Equations '7-10 are replaced by the following:
wnere Eli. represents the potential drop across the grid leak during a maximum of the audio signal, and A represents the corresponding drop when the audio signal has zero value. It will benoted that now a very wide latitude is possible in selecting Y, Z and PI since it is now only necessary that Z exceed Y (instead of that it exceed ZY).
Since Erl is independent of A, and since A is preferably small, it is convenient to assume A= for preliminary calculations. Then measure or compute K which is -the ratio of the grid current when ESI :0 and the grid leak resistor 0 to the .grid current when Esi has its preliminary value and the grid leak is adjusted to give the preliminary value of EL. Then using A=KEIL the preliminary values of Esl, Ebl and EL may be corrected.
Thev relations above described ensure the attainment of approximately linear modulation while constantly maintaining both tubes in an efficient state of` class C operation at all times, and enable the attainment of `high efficiencies at or above normal carrier level. To secure very high efficiency X, Y and Z should be kept smallsay below .'75 and preferably below .55. For high power output X, Yy and Z should be large enough so that the corresponding pulse lengths permit full 'power output without excessive current densities. Preferably the pulse lengths for X, Yand Z should each exceed 60. n
Approximately equal power division between tubes I and 2 at peak `output is desirable` and is easily attained in cases where the tubes are worked at full emission current by making X and Z roughly equal if the tubes arealike. Approximate linearity of modulation inherently results from the specified relations, lespecially if Y and Z are so related that tube I will deliver twice as much power with a pulse length of arc cosine (l-Zl as with a pulse length of arc" cosine (l-Y) without substantial variation of Et. By slightly adjusting the exact values of the various grid potentials within the range of latitude permitted by the above relations a very high degree of linearity can be readily attained as shown in curve d of Fig. 3. l
A preferred circuitfembodying my invention is shown in Fig. 2. Here the 90 phase shift producedby the quarter-wave line 3 in the plate circuit is compensated for by an equal and opposite 'phase shift produced by a quarter-wave line 5 of opposite type in the grid circuit of tube I. The quarter-wave lines 3 and 5 have for their shunt reactances, parallel resonant circuits CI-LI, C2--L2, C4--L4 and C`5-L5 detuned to give the required reactances.
It will te. noted that this circuit, incorporates both the poor`- regulation feature land the'part-ia'l grid leak bias feature above discussed. Leak-resistor RI provides part of the bias voltage 4for tube I. Also the impedance of grid line 5y with its terminating resistance R2 isy sufficiently-highA to cause voltage Erl to be greatly `reduced by grid current. 1
The modulating voltage is applied to thegrids of tubes I and 2 in the same phase, but with different magnitudes, by means of the tapped transformer T. With many types of triodesEsZ may be betweenone-half and two-thirds of Esl. The voltages Erl and E`r2'may. belequal `or un-z equal according to the adjustment of the resistance R2 terminating the grid phasing line 5.
This resistance may or may not be required,depending on how much it is desired to let the grids go positive, and the `characteristic impedance of` the grid line 5. l
. .If `the tubes are of the type known `as 2A3, approximately the'fol-lowing values may be used in Fig. 2 for 1000 K. C.: C4-L4 and C`5-L5 are tuned to give an impedance corresponding to 2,50 ,aI-I; R2=1750w; R,1=50,(l00t.v;l Esl=120 volts;
volts. Such a modulator using approximately these values in the circuit of Fig. 2,has been operated giving the characteristics shown in Fig. 3. l
Of the remaining circuit elements CB is the series element of the grid line 5 and its reactance is made equal to the desired characteristic impedance of this line. L3 is the series element of l the impedance inverting line 3 and corresponds to the inductance L in Fig. l.vv CnI and -C712, are theneutralizing condensers. .AI-A4 are direct` current meters for plate. and grid currents, BI-B are blocking condensers having negligible impedance atl carrier frequency butserving to block direct currents, and in the case of condensers BI, B2,.B3 and B5 also serving tocblock audio frequencies.
The grid line 5 should be adjusted to havea phase shift of exactly degrees. A cathode-ray oscillograph could be used by `connecting onefpair of plates across each vend of ,the network uand adjusting. for acircle on the screen, but the practice of this method requires a certain amount of skill because of the presence of harmonics in the voltages. Alsothe capacitance of the oscillograph disturbs conditions unless it is to beleft permanently connected in the circuit. A better method, which does not require the connection of any apparatus to the radio-frequency circuits, is to couple the radio-frequency input terminals TI, T2 loosely VIto the driver stage DS, after having xed C6 at the desired value as explained above, 4and then` vary Cll while watching the plate-current meter (not shown) of the driver tube. A maximum will be passed through at series yresonance between C6Y and 1.404, which is the `correct adjustment. `Before the final adjustment of C4 `is made the neutralizing condensers Cul and (In-2` should be adjusted, which `can be done by any-of the -conventional methods. CELS should also be adjusted to resonance, but is .not critical as it does not affect the relative phases of the grid voltages.` i f VThe radio-frequency excitation voltages on the grids of the two tubes I and 2 are preferably adjusted to the calculated values without connecting any measuring device, such as a vacuum-tube voltmeter, which might have 'suflici'ent capacitance to change the tuning and disturb the voltages -being measured. This can be ac` complished by using the'power tubes I and 2 as vacuum-tube voltmeters. To do this the direct-voltage supply to the plates of I and 2 is disconnected, the radio-frequency voltage is applied to the input terminals TI and T2, and the negative grid bias on each tube increased until grid current just ceases to ow. 'Ihe peak radiofrequency voltages are then equal to the bias potentials.
In adjusting the impedance-inverting network 3 there are two separate conditions which must be fulfilled. The rst is that impedances must be accurately inverted (i. e., the line must be a quarter-wave line), while the second condition is that the characteristic impedance must also be correct.
When the end sections are properly adjusted, the characteristic impedance is determined by the reactance of L3. However, the equivalent series reactance of L3 will not be the value computed from the inductance, but will be somewhat greater because of partial resonance with the distributed capacitance of the coil and wiring. Some method is hence needed for experimentally determining the proper value of L3 under actual conditions. A cathode-ray tube could theoretically be used, but has the disadvantage of introducing capacitance unless it be left permanently in circuit. Small harmonic voltages across the tank circuits also cause trouble when a cathode-ray tube is used. A simpler and more satisfactory technique is as follows:
A load resistance of the proper Value is rst connected directly across the output L2C2. If the value previously computed as correct for each tube at peak level is denoted by R, the amount of resistance used is R/2. The lament of tube 2 is then opened, but this tube is left in the socket so that its capacitance will be present. Since the tube has been neutralized, the tube capacitances will be the same with the filament cold as when it is operating. The normal value of radio-frequency grid excitation is applied, and the direct grid voltage of tube I set at any convenient Value. Plate voltage is applied, preferably of reduced value or through a protective resistance. Condenser CI is then adjusted for minimum plate current of tube I as in any class C amplier, while C2 is adjusted for maximum plate current. It will be found that the setting of CI for minimum plate current depends on the setting of C2, and that the setting of C2 for maximum plate current depends on the setting of CI, so that the adjustments are not independent of each other. It can be `shown that the correct condition for quarter-wave operation of the line 3 willV be'obtained if the adjustments are made so as to give the highest minimum plate current of tube I or lowest maximum plate current of tube I. 'I'his may be explained in more detail as follows. If C2 is left xed temporarily and CI is varied, a minimum will be passed through. If C2 is then shifted a small amount and CI again varied, a minimum is again passed. This second minimum will, in general, be at a dierent capacitance on CI and will be a diierent value of direct plate current. may be changed in steps of one or two divisions while varying CI slowly back and forth until the particular combination giving the highest possible minimum is found, which is the correct combination to use. In carrying out this procedure the minimum for each setting of C2 should be obtained and recorded. It is not desir- CII able to adjust CI and C2 alternately to get the highest minimum as the adjustments then diverge away from the correct value.
As an alternative method, CI can be changed in steps and C2 continuously varied, when a. series of maxima will be observed for the plate current of tube I. The correct tuning in this case is the value of CI which gives the lowest possible maximum as C2 is Varied. Both ways will arrive at the same Values of capacitance.
The above tuning procedure insures that the impedance inverter 3 is working properly, but it does not verify the correctness of its characteristic impedance. The characteristic impedance can, however, be determined by shifting the load from one end of the inverter to the other as follows: With the filament of tube 2 opened and that of tube I closed, and a load of R/2 across tube 2, the condensers CI and C2 are tuned as explained above and the plate current of tube I is noted. Then the series element L3 is opened and a resistance 2R is placed directly across LICI, which is retuned for minimum plate currentin tube I. The minimum plate current should then have the same Value as before. If it does not the inductance of L3 was wrong. L3 may then be changed slightly, and the process repeated until R/2 across the output of the inverter draws the same plate current as 2R across the tube directly.
As a final check on the operating conditions, R/2 is replaced across the output, both filaments are closed, and the calculated values of radiofrequency grid excitation, direct plate voltage, and grid bias are applied. At -the positive peak of modulation voltage both tubes should then have the same direct plate current and radiofrequency plate voltage and should deliver the same power. At carrier level, tube I should be passing approximately one-half as much plate current as at peak level, and tube 2 should be just starting to pass plate current, as shown in Fig. 3. At the trough of modulation, tube I should be just starting to pass plate current. Transformer T in Fig. 2 may be replaced by a tapped resistor and variable direct voltage for the purpose of making these adjustments. If the plate currents are not equal at positive modulation peaks, the grid leak RI may be changed, or the ratio of radio-frequency grid voltages can be changed slightly by adjusting R2. Then, as modulating voltage is applied, the radio-frequency load current should change linearly. Finally R/2 is replaced by the antenna or other desired load by coupling the antenna to the tank circuit L2C2.
In the above discussion of the effect of the socalled quarter-wave lines in the plate and grid circuits, it was assumed that these networks actually acted as quarter-wave lines at the frequency applied to them. This property, in the plate-circuit network for example, depends upon the reactances of the series and shunt elements of the line. Since the reactance of inductances and parallel-tuned circuits varies with frequency, the correct conditions for impedance inversion may not be exactly fullled at the higher sideband frequencies. This is somewhat analogous to an ordinary class C amplifier in which the tank circuit is not quite correctly tuned for the side bands of high modulation frequencies. In some cases in a high-quality system it may be desirable to construct a network which is more nearly the true electrical equivalent of a transmission line, as for example by using two or three smaller sections connected end to end instead of one large section. At the higher frequencies an actual physical transmission line could be used,
The several curves of Fig. 3 respectively illustrate the important operating characteristics of the system represented in Fig. 2, when used with small receiver type triodes of the 2A3 type and when adjusted as above "outlined," In curves a,
b, c and d, the platel currents of tubes I and 2, the carrier frequency output current of the whole system as measured in load 4, and the power efciency ofthe complete output stage are respectively plotted against the instantaneous modulating voltage ESI.
Curve d shows that even with small receiver type tubes and correspondingly low plate potentials and powers the momentary eciency is above 60% not only in the unmodulated condition but also throughout the modulation cycle provided the amplitude modulating signal Ecl does not exceed volts (corresponding to about 18% modulation as seen from curve c) Furthermore, even during periods of 50% modulation it will be seen that the momentary efficiency remains above 60% during about two-thirds or three-fifths of eachl audio cycle and remains above 50% during almost the whole of each cycle. The average eiciency even at 50% modulation is therefore well above 50%. At greater modulation percentages the average efficiency is again higher because the power handled during modulation peaks so far outweighs the power handled during modulation troughs that the relatively high e'ciency at the peaks becomes the controlling factor which principally determines the average efficiency at high percentages of modulation.
Curve c is evidently a substantially straight line, thus showing clearly the practical linearity of the modulation envelope with respect to the modulating audio signal Although certain embodiments of my invention have been described in detail to enable anyone skilled in the art to construct and operate the same many variations are possible. For example, the radio frequency and audio frequency signals applied to the grids may be a single cornplex voltage instead of two independent voltages. Also, though it is preferred to bias tube 2 -so that the predetermined signal level at which it begins to operate is equal to zero audio signal and corresponds to normal carrier level (i. e. one-half the carrier amplitude of peak output) it is possible to displace this level so that tube 2 comes into operation at a somewhat higher or lower level. In such a case the powers and/ or the anode supply voltages for the two tubes may differ. In fact such unequal power division and/or supply voltages may be employed even when tube 2 comes into operation at the half-way point of the audio signal as described.
Broadly my invention contemplates grid modulation and class C operation of tube I in a circuit of the type wherein tube 2 is inoperative at low levels and wherein by virtue of impedance inversion the effective load on tube I is sharply reduced when this tube 2 does come into action. So far as I am aware it has not heretofore been known that grid modulation could be applied to such a tube as tube I, whose effective load is subject to discontinuous variation, Furthermore, so far as I am aware it has not hitherto been fknown that class C operation was possible with such a tube nor how such class C operation could be attained. The fact that both grid modulation and class C operation could be used in such a system and the further fact that such a system is inherently roughly linear and can readily be made accurately linear are, so far as I am aware, totally unexpected.
Although certain specific embodiments of my invention have been shown and described for the purpose of illustration, it will be understood that adaptations, variations, and modifications thereof occurring to one skilled in the art may be made without departing from the spirit of my invention as dened in the appended claims.
What I claim is:
1. A system for delivering modulated carrierwave power to a load, which comprises a iirst and a second discharge tube each having a cathode, a grid and an anode, means for supplying continuous potential to said anodes, connections for 4 delivering power from the anodes of both of said tubes to said load, an impedance inverting network connected between the anode of one of said of said first discharge tube an impedance whose value decreases with increasing power output' from said second discharge tube, means for applying carrier frequency signals to the grids of both said tubes, means for applying modulation frequency signals to the grids of both said tubes, means for biasing said second tube to be inoperative at all values of said modulation signal below a predetermined value and to operate in class C at all values of said modulation signal above said predetermined value, and means for biasing said rst tube for class C operation at all values of said modulation signal.
2. A system for delivering modulatedcarrierwave power to a load, which'comprises a iirst and a'second discharge tube each having a cathode, a grid and an anode, means for supplying continuous potential to said anodes, connections for delivering power from the anodes of both of said tubes to said load, an impedance inverting network connected between the anod-e of one of said tubes and said load so as to present to the anode of said first discharge tube an impedance whose value decreases with increasing power output from said second discharge tube, means for applying carrier and modulation frequency signals to the grids of both said tubes, and means for establishing a substantially linear characteristic, comprising means for applying to the grid of said first tube a bias potential sufficient tovjust cut ofi the flow of current at the most negative value of said modulation signals and to cause said first tube to pass current in short pulses of less than t6 of a cycle duration during intervals of zero value of said modulation signals, and means for applying to the grid of said second tube a bias potential sufficient to just cut off the flow of current at zero value of said modulation signals and to cause said second tube to pass current in pulses less than M2 cycle ,long and of such duration'that said second tube has a power output substantially equal to that of said first tube during intervals of maximum positive value of said modulation signals.
3. A system for delivering modulated carrierwave power to a load, which comprises a first and a second discharge tube each having a cathode, a grid and an anode, means for supplying continuous potential to said anodes, connections for delivering power from the anodes of both of said tubes to said load, an impedance inverting network connected between the anode of said first tube and said load so as .to present to the anode of said first discharge tube an impedance whose value decreases with increasing power output from said second discharge tube, means for applying carrier and modulation frequency signals to the grids of both said tubes, and means for establishing a substantially linear characteristic, comprising means for applying to the grid of said first tube a bias potential sufficient to just cut off the ow of current at the most negative value of said modulation signals and to cause said first tube to pass current in pulses whose duration during intervals of maximum value of said modulation signals is between 1.1 and 1.5 times their corresponding duration during intervals of zero value of said modulation signals, and means for applying to the grid of said second tube a bias potential sufficient to just cut oi the flow of current at zero value of said modulation` signals and to cause said second tube to pass current in pulses less than 1/2 cycle long and of such duration that said second tube has a power output substantially equal to that of said first tube during intervals of maximum positive Value of said modulation signals.
4. A system for delivering modulated carrierwave power to a load according to claim 1, wherein said means for biasing said first tube comprises a condenser and a grid leak resistor connected to produce a bias Voltage in response to the flow of grid current in said rst tube.
5. A system for delivering modulated carrierwave power to a load according to claim 1, wherein said means for applying carrier frequency signals to the grids of both said tubes includes a source of carrier frequency signals hav- V ing an impedance sulciently high to cause the voltage of said carrier frequency signals applied to the grid of said first tube to fall oi substantially in magnitude responsive to the flow of grid current in said tube.
6. A system for delivering modulated carrierwave power according to claim l, wherein said means for applying carrier frequency signals to the grids of both said tubes include a source of carrier frequency signals, a phase shifting network, connections from said source of carrier wave signals directly to the grid of said second tube, and connections from said source of carrier frequency signals through said phase shifting network to the grid of said first tube, the effective impedance of said phase shifting network and source as seen from the grid of said second tube being sufficiently high to cause substantial falling off of the voltage of said carrier frequency signals on the grid of said rst tube in response to the ow of grid current in said rst tube, and wherein said means for biasing said first tube include a grid leak resistor and a condenser for providing a bias potential responsive to the flow of grid current.
7. A system for delivering modulated carrier- Wave power to a load according to claim 2, wherein the carrier frequency, the modulation frequency and bias voltages applied to ythe grid of said first tube are equal respectively to 1 Et -ZPl "t'i Y EL ZPl l-- -l- A and wherein a grid leak is also connected to the grid of the rst tube to provide a further grid bias voltage whose value during intervals of maximum power output is equal to where al is the amplification factor of the rst tube, A is the value of said further grid bias voltage provided by the grid leak during intervals of normal output, P1 has a value of the order of 1/5 of the anode supply voltage divided by 1, Et has a value of the order of 4K5 the anode supply voltage, Y has a value between .25 and .75, and Z has a value between .40 and .90.
8. A system, for delivering modulated carrierwave power to a load according to claim 1, wherein said means for applying carrier frequency signals to the grids of both said tubes includes a source of carrier frequency signals and a high impedance between said source and the grid of said rst tube.
FREDERICK E. TERMAN.
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
BE441445D BE441445A (en) | 1939-06-10 | ||
BE468893D BE468893A (en) | 1939-06-10 | ||
US278526A US2238236A (en) | 1939-06-10 | 1939-06-10 | Modulation system |
FR867663D FR867663A (en) | 1939-06-10 | 1940-11-08 | Modulator systems |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US278526A US2238236A (en) | 1939-06-10 | 1939-06-10 | Modulation system |
Publications (1)
Publication Number | Publication Date |
---|---|
US2238236A true US2238236A (en) | 1941-04-15 |
Family
ID=23065316
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US278526A Expired - Lifetime US2238236A (en) | 1939-06-10 | 1939-06-10 | Modulation system |
Country Status (3)
Country | Link |
---|---|
US (1) | US2238236A (en) |
BE (2) | BE441445A (en) |
FR (1) | FR867663A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US2487212A (en) * | 1946-06-19 | 1949-11-08 | Zenith Radio Corp | High efficiency modulator |
US2616075A (en) * | 1945-06-16 | 1952-10-28 | Rca Corp | Signal voltage frequency converter |
US2658959A (en) * | 1949-11-02 | 1953-11-10 | Bell Telephone Labor Inc | High efficiency radio-frequency power amplifier |
US3114107A (en) * | 1960-12-06 | 1963-12-10 | Packard Bell Electronics Corp | Radio frequency transmitter |
-
0
- BE BE468893D patent/BE468893A/xx unknown
- BE BE441445D patent/BE441445A/xx unknown
-
1939
- 1939-06-10 US US278526A patent/US2238236A/en not_active Expired - Lifetime
-
1940
- 1940-11-08 FR FR867663D patent/FR867663A/en not_active Expired
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US2616075A (en) * | 1945-06-16 | 1952-10-28 | Rca Corp | Signal voltage frequency converter |
US2487212A (en) * | 1946-06-19 | 1949-11-08 | Zenith Radio Corp | High efficiency modulator |
US2658959A (en) * | 1949-11-02 | 1953-11-10 | Bell Telephone Labor Inc | High efficiency radio-frequency power amplifier |
US3114107A (en) * | 1960-12-06 | 1963-12-10 | Packard Bell Electronics Corp | Radio frequency transmitter |
Also Published As
Publication number | Publication date |
---|---|
BE441445A (en) | |
BE468893A (en) | |
FR867663A (en) | 1941-11-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US2220201A (en) | Modulation | |
US2294800A (en) | Modulation system | |
US2326314A (en) | Frequency modulation | |
US2211003A (en) | Radio signaling system | |
US2324279A (en) | Amplifier | |
US2238236A (en) | Modulation system | |
US2516181A (en) | High-power audio frequency amplifier | |
US2347458A (en) | Frequency modulation system | |
US2777020A (en) | Direct coupled high fidelity amplifier | |
US2658959A (en) | High efficiency radio-frequency power amplifier | |
US2498711A (en) | High-frequency amplifier | |
US2255476A (en) | High efficiency amplifier | |
US2006872A (en) | Series oscillator | |
US2036164A (en) | Phase modulation | |
US2235677A (en) | Amplifier for signal transmission | |
US2136479A (en) | Modulation system | |
US2279661A (en) | Wave control and control circuit | |
US2419793A (en) | Ultra high frequency electron discharge device circuit | |
US2241892A (en) | Thermionic amplifier | |
US2041951A (en) | Modulating system | |
US2331821A (en) | Frequency modulation | |
US2045973A (en) | Modulation | |
US2026944A (en) | Means for receiving and amplifying electric signals | |
US2763734A (en) | Push-pull circuit for amplifying or producing high-frequency oscillations | |
US2154783A (en) | Electrical transmission system |