US2253976A - Electrical oscillation translating system - Google Patents

Description

l. w @GUANELLA ELECTRICAL oscILLmIONs TRANSLATING sYsT'Eu v Filed Dec'. 7.' 1939 sheets-sheet 2 I INVENTOR Gus'awuanellgiv- ATTORNEY Filed Dec. 7, 1959 3 Sheets-Sheet 3 oY 024- l 23 INVENToR* Gusawuan eZZa.

ATTORNEY nPatented Aug. 26, 1941 l UNITED" STATES PATENT OFFICE New York Application December l, 19.39, Serial No. 308,!54

In Switzerland December 2, 1938 14 owns. -(el. 11s-1u) I This invention relates to electrical translation systems, such as amplifiers or converters of os cillations. In particular, the invention is concerned with electrical amplifiers, electro-acoustical converters, radioe and other transmitters or receivers of oscillations or bands of oscillations having frequencies of any order or range from low- `or audio-frequency up to the highest opposition with the input energy so as to cancel as far as possible undesired distortions and disturbances present in the output energy and to secure a linear relation between the output and input magnitudes. By such an inverse feedback the gain of the amplifier or the amplitude of the translated magnitude was reduced and necessitated additional amplification in order to com- Y pensate for the gain' reducing effect of the inverse feedback. Furthermore, undesired phase? cuide substantial distortions of the ampnned energy.

It is another object of 'the invention to prevent the building up of oscillations in a translating system provided with inverse feed back to a critical point where self-sustained oscillations are developed.

Itis still another object of the invention to compensate for disturbances of any kind and nature arising in an amplifier system tending to instability. such as due to the presence of feedback.

It is still another object of the invention to translate relatively wide frequency bands without distortions, particularly self-generation at extreme frequencies of the band due to feedback.

According to the present invention, the effective output magnitude and input magnitude,

or magnitudes proportional to them,l are .com-

pared by means of feeding back a Vfraction of that output magnitude to the input and com-- bining it with the input magnltudegfor part thereof, in opposite phase relation, and developinga correction magnitude varying in dependence upon the combined magnitude which later is caused to act recurrently or vperiodically upon the input of the amplifier so as to reduce. or.

cancel, substantially distortions' and other disturbances or deviations from the predetermined.

' such as linea gain of the magnitude undershifts of the translated oscillations. particularly of the extreme frequencies of a band of oscillations being fed back caused even a continuous increase of the amplitude, instead ofa decrease, and undesired self-generation of sustained oscillations. In vmore complicated translating systems aneffective inverse feedback permitted therefore the translation of very narrow wave bands only, or it necessitated complicated and expensive additional measures to prevent instability conditions or undesired self-generation of oscillations. f

It is an object-of the invention to -do away with the above drawbacks of known translation systems comprising aninverse feedback, anddto substantially eliminate instability conditions.

Itis a further object of the invention to imi` prove a translation system provided with inversev is proportional to the efr going translation.

According to a particular feature of the in, ven-tion, the oscillating input energy, or part thereof, is recurrently or periodically interrupted, such interruptions occurring before undesired 'disturbances can develop or persist to an unde- "slred extent. During the periods where the input energy is permitted to' pass developed by and depending on-a deviation of hthe input from the effective output magnitude which are compared by means of an inverse feedback arrangement. A, l

For comparing .the fed back tude which tive output magnitude.

with the incoming magnitudefa control circuit is feedback as to the` gain obtainable with a given c tem provided with inverse feed back and to exprovided by the invention which develops correcting magnitudes 'varying' substantially in dependence upon a deviation, if present, between the compared input, and edectiv'e output. magni- 'tudes from a predetermined translation law, and

` these correcting fmagnitudes developed bythe control circuit are recurrentiy applied to the in- `put of the amplifier or translation system to the the translation system, a correcting magnitude is added thereto effect of reducing or removing that deviation. The comparison between the input and output magnitudes in the control circuit is made periodically or recurrently at a frequency considerably exceeding the frequency of the input oscillations, so that sudden as well as continuous changes of the output magnitude can be made successively and within the shortest periods to correct the input magnitude; thereby the effective outputy ,of the translated energy and disturbances of the incoming energy can be effectively suppressed, entirely or to a great extent without encountering other drawbacks, such as instabilities or selfgeneration effects.

'Ihe correcting magnitude is developed and changed recurrently or periodically by the controlling circuit, ,but retains substantially its instantaneous value between two successive changes which occur during the short periods while the oscillations pass the system. The variations of the correcting magnitude are thereby substantially proportional to the effective input energy of the translation system which equals substantially the incoming energy reduced by the fed back energy applied thereto in phase opposition.

'I'he invention will be more clearly understood and its above outlined and other objects become more apparent when the specification proceeds with reference to the drawings.

Fig. 1 shows in a schematical way the principles of operation of the known inverse feedback arrangements, Fig. 2 a time-amplitude diagram of operation of an arrangement according to Fig.

1, Fig. 3 in a schematical way the principles of d operation of the invention, Fig. 4 a time-amplitude diagram of the operation of an arrangement according to Fig. 3 upon reception of a starting impulse, Fig. 5 a similar time-amplitude diagram showing the conditions arising according to the invention if a variable magnitude is received, Fig. 6 a time-amplitude diagram similar to that of matical way a known inverse feedback system, comprising a translation system A. such as an amplifier and a feedback channel B. 'I'he ampliiication factor or gain a of the translation system A is normally complex; the feedback channel B may have an amplification factor or gain b. .P is the input magnitude; R. is the magnitude fed back by channel B. R and P are impressed together upon the input of A, so that the effective output magnitude Q of the translation system A answers the following condition:

The amplitude amplication factor or gain Q/P of the entire translation system decreases therefore with increasing inverse feed back below a and approaches eventually the value l/b. Simultaneously the distortions are reduced. The

'relative amount on second harmonics is reduced the rapidly increasing compensating magnitude R to the value given in (1).

Fig. 5 showing the conditions if a correcting mag- 1 nitude is recurrently applied, Fig. '1 in a schematical way the principles of a modification of the invention, Fig. 8 in a diagrammatic way a correcting circuit suitable for practicing the invention, Fig. 9'in a/diagrammatic way the embodiment of such a correcting circuit in an amplifier system, Fig. 10 in a diagrammatic way a circuit arrangement embodying the invention for modulating optical rays, and. Fig. 11 in a diagrammatical Way a transmitter circuit embodying the invention. Identical reference characters in the drawings denote like parts or devices.

It is to be understood that the invention is not limited to any of the exempiiiications shown in the drawings but to be. derived in its broadest aspect from the appended claims.

Referring to Fig. 1, there is shown in a sche- According to the known investigations of H. Nyquist (Bell System Technical Journal, 1932, p. 126) a self-generation of oscillations becomes possible in most of the cases where the complex amplification factor a.b of the circuit becomes less than 1. Inl multistage amplifiers with inverse feedback the phase shifts depending on frequency in the individual stages usually result in an absolute value of a.b which is large compared with 1, and consequently an efficient inverse feedback cannot be easily applied.

Fig. 3 shows the operation of a recurring, preferably periodically recurring linearisation or stabilization according to the invention which prevents undesired self-generation or other instabilities of the system. A is again a translation system to be linearised or stabilized, and B is a channel for feeding back a fraction R of the output magnitude Q. At the end of each perlodically recurring period To the instantaneous difference between the input value Pm and the fed back value Rm at the moment tm=m.T is converted by device C into a corresponding change sm of the effective input value S impressed upon jthe translation system proper A. Therefore A gaan-R1.)- 2) wherein c is a constant. The magnitude S at the moment immediately following tn amounts therefore to The period Ts between two subsequent changes s may now be assumed long compared with the total of the time constant T. of system A andv the mostly negligible time constant Tt of feedback channel B so that the eiIeetive output mag- 1nitude Qn in the moment tn can always be writ- The' constants a, b, care to be chosen gous way, and the output magnitude Qn is at the Assuming areal translation factor or gain. of the feedback channel B, it follows and for the correcting magnitude according to (2) follows Since analogous relations exist also for Qn-i, Qs-z, it results for P-1=0 and Q=0:

condition is answered as completely as possible, whereby j'.

in any case is a fraction smaller than 1. Then the total Pn-x approaches after a few periods To ifP can be assumed to be constant during that time, and for Q results:

` short period of time.

Referring to Fig. 4 it may be -assumed that fthe input magnitude P is formed by a continuous impulse U. The magnitudes P, Q, R, S are zero until impulse U is received. This impulse U causes at the moment ti a first increase of the effective input magnitude S of thetranslation system A according to (2) :i

' The period Tc until the next change of S occurs may be chosen to be considerably longer than. the time constant Tr. of system A so that the 'effective output magnitude Q already moment tz assumes the value which because of (9) differs only'sli'ghtly'from the final output magnitude Q=1/b-U.} In the moment t: the instantaneous input magnitude P2 diners from the fed back fractionjRz of the l output magnitude Q2 only by the amount (P2-R2) =UbQ2=.-f U which causes anew change sncf S by the amount of s2=cfU whereby according to (8) results:

e.=vaS.=aa+s2 f-% 1r v v 12) Thisphenomenon repeats itself in an analoso that the -'e'xactly to the nih change of the correcting magnitude S as can also be derived from (8) Since f is a small magnitude according to (9), fm rapidly approaches zero with increasing index so that Pn-1I=fn1-U disappears and Q rapidly reaches the exact final value Q= U/b.'

This rapidy and exact equalization of the effective output magnitude toits theoretical value in the absence of feedback forms an essential characteristic of the present invention. If the constants are properly chosen, the effective output magnitude according to (11) after a' single period Tc will equal approximately the changed theoretical value, and after the following second period Te the equalization according to (12) is very exact. Larger deviations of the constants from condition f (9) result solely in a longer period vof time until the final value Q is reached without impairing its accuracy.

In Fig. 5 the conditions are shown with respect to-a continuously changing or varying incoming value P. The change or variation s3=o(Pa-bQa) of the correcting magnitude S atte comprises,for instance, a first portion -tude results which at any time conforms very theoretical value in absence of a feedback, if a somewhat inaccurate translation of the preceding change of the theoretical value i progress of Q only the frequencies wp before the l due to the choice of c is disregarded.- At the moments t1,` tz, Qn corresponds with the' value given by (8) which always deviates by the small amount P11-1 only from the theoretical value.

The Fourier analysis shows that in the stepwise comprised by P and the frequencies (wciwp), (2wpztwp). (3wcwp), are contained wherein we is the change in frequency within the period Tc. Therefore, if we is chosen large or high enough these higher frequencies can vbe suppressed, if desired, in a simple way by means of low-pass filters so that only the original frequencies wp remain the progress of which is shown by ments, building up of self-oscillations is entirely excluded in system A even with uncontrollable frequencies as long as Ts exceeds T. and f=(1-abc) is a fraction smaller than unity. This is due to the fact that any disturbance of the system which could result' in self-oscillations or other instabilities is continuously and in rapid sequence reduced or quenched by the periodically recurring changes of the correcting magnitudes so that. the eilective output magnitude always tends towards the theoretical value prescribed by the instantaneous input magnitude.

' Whereas withy known inverse feedback arrangements the faults of translation are reduced only and cause according to (1) a corresponding reduction ofthe gain in amplitude, the effective Some faults in translation may result only from changes in P during the last periods Tc, if

condition (9) is not answered satisfactorily..

Such deviations can be completely corrected afterwards by corresponding corrective changes of the controlling magnitude.

The changing or correcting frequency we should normally be chosen as high as possible so that also at the highest frequencies to be translated, i. e. at fastest changes of P, the fault p in the translation remains sulciently small according to (8) and the frequencies we and wp in the effective output magnitude Q can be separated cleanly. Therefore, we advantageously exceeds 2wp. With a very large we, however, period Te may become smaller than time rate T. of system A so that (4) does not apply anymore. By choosing a sufllclently small constant c and abandoning the exact observation of condition A(9'), nevertheless, self-generation or other instabilities can still be safely avoided. However, Q does not approach the changed theoretical value so fast anymorepas is the case with progressive reduction of faults according to (13).

It has been assumed above that the changes in the correcting magnitude S after each period Tc occur suddenly and depending on the instantaneous error of Q. 'Ihese changes can be obtained, according to the invention, also continuously during predetermined periods Tl. This is shown in Fig. 6, where again a translation is shown similar to that of Fig. 5. It is assumed that the entire change of S within a given period occurs accordingv to (2). The operation results ink a progressive reduction of all faults of the effective output magnitude Q. as can be'learned from an inspection of Fig. 6 without further ex-v planation, if those given with reference to Figs. 4 and 5 are taken into consideration.

In an arrangement according to Fig. 3 the correcting magnitude S is recurrently changed depending on P. The magnitude P, however, can also be directly introduced into system A as shown in Fig. 7, and the correcting magnitude S operates only to correct errors in the translation. To this effect a periodically or recurrently operating control device or circuit C may be arranged in series with network D which has the Same translation characteristics as the series arrangement f A and B. Consequently the time constant Td and amplification factor or gain d of the network is:

has been reached. `This final effective vvalue is arrived at most rapidly if 15) is actly as possible.

'I'he advantage of such an arrangement lies therein that the output magnitude is produced by continuous and steady translation of the input value through A within the smallest time delay T. whereby only the instantaneous errors of Q cause a small additional and periodically recurring change of S. The stepwise changing component of Q is therefore ofv a very small amplitude and the time of translation through the entire linearized system is a minimum.

A may be, for instance, an 'electrical translation system, such as an amplifier to be linearized. In such a case the magnitudes P and Q represent electrical currents or voltages. The invention fulfilled as exmay also be applied to'converter systems, such as e. g. electro-acoustical transformers. In the latter. case, system A may be a loudspeaker which convertsy incoming electrical into .acoustical oscillations while B is a controlling microphone which translates the acoustical oscillations Q reproduced by A into electrical oscillations R which are compared in C withl the incoming electrical oscillations P. If, however, A is a system for convertlng incoming electrical oscillations into variations of the intensity of light oscillations, then the light variations Q are translated into electrical phenomena R by means of a light-responsive electrical system B. System A, however, may also serve to modulate a carrier wave with incoming electrical currents or voltages so that the output magnitude Q represents a momentary or instantaneous state of modulation of the carrier wave controlled e. g. by the low-frequency electrical magnitude P in linear dependency. Then the feedback system B represents a controlling receiver which reproduces the low-frequency magnitude by demodulation.

In all the above cases, which are to be considered as exempliflcations only, the recurring` linearization results in a relation between P and Q determined according to (10) and (16), respectively, solely by the constant b of system B, or the constants b and d of the systems B and D. Since the translation characteristics of these systems can easily be made constant and free of distortion. the inevitable disturbances, distortions and variations in translation of system A can be substantially eliminated. l

Referring to Fig. 8, a correcting circuit C embodying the invention is shown. The incoming oscillation P and the fed back oscillations R are impressed upon the transformer Le, L1 through terminals I, 2 and 9, I0, respectively, in such a way that they counteract each other. A periodical or recurrent controlling voltage X of an auxiliary generator E is derived from the terminals 5,y 6 and applied through center taps of coils L1 and L2 to the rectifiers or gates G1, G2 arranged in push-pull. If this controlling voltage'is negative, i. e. acts in the cut-off direction of the rectiers, the latter cannot be passed by current, even if there is a relatively smallvyltage difference between the ends of coil L1 resulting from a difference of the inputvoltage P and the inversely acting voltage R. As soon as the controlling voltage becomes positive, i. e. acts in the conducting or pass-direction of the rectiers, a corresponding current flows through the centers of the coils L1, Lz and simultaneously through the rectifiers which now can be passed also by the currents resulting from additional voltage differences (P-R) if prevailing between the 'ends of coil Li,

. tween the end of coil Le.

I potentials at its ends so impulses correcting circuit C is the duration or period f an'ampliiler A1 is shown. The incoming signals accelera so that corresponding impulse potential differences appear at the ends oi coil L.- and are impressed upon an ,integrator circuit comprising a pair of series resistors W1 and W2 and a parallel condenser adapted to produce an output voltage S across 3, l whose changes correspond to the magnitudes ot the impulse potentials appearingbe- The correctingqvoltage tap of coil In causes equal that no current is passed thereby through the resistances W1 and W1.

y Thus by (P-R) a charging current X applied to the center Wi'i-Wz is passed through the resistancesiWu W1 to condenser K1. Consequently the voltage S on the capacity K1 changes during period T while the rectiiiers are alive, by an amount A :i MFL- El.: K1T' K1(W.+W.) c-(P', R) im As soon as X becomes negative, the charging current is interrupted and the voltage S remains unchanged until X changes its`direction again. T hus voltage S ischanged after each recurring period T`by an amount s which is proportional to the instantaneous voltage difference P-R while a short switching-in impulse X prevails.

The controlling voltage X can be produced, ior instance, by means of a generator E comprising a condenser Kn which is charged by a source H21 through a load resistance W11 until the voltage built up across the condenser suillces to cause a discharge through a gas-filled tube V21. The voltage drop caused by the discharge current alongl` resistance Wn: put on the rectiilers G1, G1 from tap Il of source Haz so that the voltage X acting simultaneously upon the rectiers becomes positive during that short discharge period. In order to prevent an undue increase of X'during the discharge, a recti' iler G21 can be provided which is passed by current when its biasing voltage tapped at I3 from source H21 is surpassed,A whereby a maximum limit for the voltage X is set. The frequency of the discharges through the Thyratron-type tube V11 can be adjustedge. g. by changing the biasing voltage on its grid tapped from H21 Land, oi' course, by adjusting the capacity Kn. f

li the positive impulses X are very short then the changes a of the"efiective input be considered as occurring suddenly. If' these are oilonger duration T., they result in steady changes of S so that linearization occurs v in a way similar to 'that'e'xplained with reference to Fig. 6. The translation constant c of the givensaccordlng to (l'll by 'I'. of the impulse andthe time constant K1(W1+Wz) it .additional voltage losses in the rectiers are neglected as is always permissible. l'

Itis to be understood that a translation channel A is connected with terminals I, 4 wh'ile a feedback channel B is connected with terminals 9, I in the way shown e. g. in Fig. 3, `terminal I being connected e. g. to'ground or the-cathode and terminal I tothe input grid o! an amplifier.

There are many other known arrangements capable of producing a suitable controlling voltage X.-y In many cases also a sinusoidal alternating voltage can be used for controlling the correcting circuit C.

In Fig. 9 the application of the invention to exceeds the biasing voltage f of a voltage P are impressed through a transformer La, L4 upon the terminals I I, I2. The fed back voltage R is derived from a tap of a voltage divider or resistance W4 in B1 and fed back through a conductor between terminals I3, I4, I2 and con L1 to the center tap of con L5 which impresses it upon the rectifiers Gs, G4 arranged back to back. 'Ihe controlling alternating voltage X is impressed through terminals 5. 6 and coil Le of controlling circuit C1 uponv coil L5 and rectiilers Gs, G4 connected in series therewith. Thereby the rectilers are made alive simultaneously when the voltage caused by X acts in their conducting or pass-direction, and they can then also be passed. by the current impressed upon terminal Ii by (P-R). Between the rectifiers G3 and .G4 again resistance W1 is connected in series with capacity K1 and a conductor leading over terminals 4, 8, or ground, to one end of voltage divider voltage S on the terminals 3, l is changed every g time by a value s corresponding to (2) The voltage resulting from X across the series arrangement of the rectiilers even in their pass-direction cannot cause, however, a flow of current through W1 and K1 because their circuit is attached to the e centre points of L5 and between Ge and G4.

By amplification in channel A1 an output voltage Q results between the terminals 1, 8 equalling a S.

It may be assumed that the ampliiier channel comprisesl four tubes V1 to V4 coupled by resistance-capacity-couplings. According to (8) the total amplification degree or gain Q/P depends with great approximation on the reduction l/b in the feedback channel B1 which can be adjusted bychanging the tap point onl voltage divider W4. If b is thus adjusted, it is also of advantage to change the gain a of channel A1 so that condition (9) is fulfilled as well as poible.

voltage S can lThis can be easily achieved e. g. by mechanically coupling the adjusting members of the voltage dividers W4 and Wa, the latter arranged in the amplification channel A1.

The arrangement described represents a voltage linearization because the controlled output ymagnitude Q is a voltage. Linearization oi.' current can be obtained if the entire output current is passed through an ohmic resistance causing a 'voltage drop R, proportional to the current and fed baci: upon the correcting circuit C1.

It is understood that an auxiliary generator e. g.-

of the type E (Fig. 8) is to be connected with the terminals t, t in Fig. 9.

Further details of the amplier etc.- which are conventional 'are not shown in the 'drawings for the sake of clearfness. f

Referring to Fig, 10, there is shown a light control system embodying the principles of the invention. It is assumed that the brightness of a source oi light A: depends upon the currents or voltages supplied by an amplier Az. Aa may bel for instance, agas-discharge lamp of wellhiown construction.v The light beam Qis projected by a reflector F and its intensity should correspond with the effective input, voltage S of the ampliiler A2. A linear dependency of these magnitudes is, however. impaired by disturbances and distortions in the amplier Ai and the electrically controlled light source Ar. A small part of the light einergl ing from' A: is directed upon the light-responsive by an amount upon the grid of tube V11.

. device or photocell Bz and causes therein proportional currents or voltages which are impressed upon the correcting circuit C2 in inverse direction to the incoming signals P.

'I'he correcting circuit C2 comprises in this embodiment of the invention electron tubes Vs, Ve,

. arranged back-to-back between points 2l, 22;

terminal I is connected through La with 2|, while terminal 2 is connected with 22 through the high ohmic impedance We. 'I'he resistance of the tubes is changed simultaneously periodically or recurrently by a controlling voltage X applied to the grids of tubes V5, Ve through transformers L20, Lv, La, so that the impedance between'the points 2l and 22 varies between a small and a practically infinitely great value. Consequently, the voltage on the controlling grid of the pentode V1 connected with 22 varies between a zero-value and the instantaneous value of P--VR developed across We and appearing at 22. The capacity K1 arranged across the plate-cathode path of tube Vv is of such great value that voltage variations at the plate of V7 resulting from the variations of the voltage on the controlling grid are very small compared with the entire plate voltage. A voltage impulse of the amplitude P-R and the .duration T applied to the controlling grid of tube V7 results therefore, according to the steep characteristic y of the tube, in a plate current impulse -1/(P-R) of equal duration, which passes condenser K1 and causes a change of its Voltage Thew correcting voltage S between the input terminals 3, t of amplier channel Az is therefore periodically or recurrently changed in the waydependent only on b. If the translation characterstics of the feed back channel comprising phtocell B2 and amplifier B: are constant and. independent of changes of amplitude, the effective output magnitude Q is alwaysl free of disturbances and distortions.

Referring to Fig. 11, there is shown an embodiment of the invention in a radio transmitter comprising an amplifier channel A: for the lowor audio-frequency oscillations S, s modulator A4 for modulation of the amplitude of the high frequency carrier waves Y, a pick-up receiver B4 for the radiated modulated carrier wave to obtain the magnitude R therefrom by demodulation, and a controlling circuit C: for comparing R with thelowor audio-frequency input magnitude P.

The modulator A4 is shown by way of example as an arrangement for modulating the plate voltage. The low frequency oscillations intro` duced through terminals 33, 34 and transformers l Ln, L14 are impressed upon the grid of tube Vis and control othe plate current o f that tube. The variationsof that current are translated through transformer Lis upon the plate voltage of tube The carrie'r wave is introduced through terminals 23, 24 and" transformer Lm and impressed The oscillating circuit Ka, Lia in series with the plate of tube Vu is tuned to the high-frequency of the carrier wave while the amplitude Q thereofis modulated by the low frequency voltage introduced through terminals J3, Il. The thus modulated high frequency carrier is impressed through transformer Liv, Lis upon the antenna attached to the terminals 21, 2l. The controlling receiver B4 comprises an input circuit Ka, Lu tuned to the frequency of the carrier and coupled with the coil Les. With the terminals IIS, |20 an antenna (not shown) is connected which picks up radiated modulated carrier'wave energy with relatively large amplitude due to its proximity to the transmitting antenna connected to 21, 28. If desired, however, an amplifier (not shown) can be inserted in series with coil Leo. In Fig. 11 two rectifiers of the type of diodes V12, Vn are arranged in parallel and their anodes connected with the same polarity to the ends of coil L20. Since these ends are in phase opposition, it appears that the rectiiiers Viz, V13 operate in push-pull. 'I'he cathodes of the rectiiiers Vu, V13 are connected at 29 and from there with terminal 30 while terminal 3l is connected with the center point of coil Lao. Between the two leads to terminals 3B, 3i a resistance Wu and capacityf'Kiz are arranged in parallel. It will be appreciated that by proper tuning and dimensioning of condenser K12 the -high frequency component of the rectified modulated carrier oscillations is /short circuited and only the rectined low frequency component passed on to the terminals 30, 3|, forming the magnitude R which includes due to the resistance W1: thel fastest low frequency oscillations of the high frequency amplitude Q.

Since the fed back low frequency magnitude R and the low frequency input magnitude P are impressed in inverse relation upon the transg voltage X is applied through terminal I in parallel to controlling grids of tubes Va, Va and to their cathodes through terminal 8.' The cathodesl are also connected to the center point of the secondary of Lu, andthe ends of the latter are connected to another grid each of tubes Va, Vs.

During the short period while P'R causes a current iiow in the way described, condenser Ki is charged `by a current caused to flow by the secondary of Ln over resistances W1, Wa, as set forth above with reference to Fig. 8, and the effective input magnitude S is recurrently built up or changed according to P-R between terminals 3, I of the ampliner channel A1 which translates thislow frequency magnitude through terminals 3l, 84 to coil Lu of transmitter A4. In

the auxiliary controlling frequency lX. To .thisv y eect a low-pass filter can be connected with the amplifier channel Air, to the effect that this auxiliary frequency and its harmonics are suppressed.

Alternatively, or in ,addition thereto, the corresponding side bands of the carrier wave Y can be filtered out by means 'of tuned circuits. In such a case it is advantageous to feed back the still non-filtered high frequency oscillations through the recter circuit B4 instead of flltered or transmitted high frequency waves; thereby the time constant T. of the fed back magnitude is not unduly increased by the period for building up y -linearizatlon and progressive reductionsof errors in the effective output magnitude.

It will be appreciated from the above explanation of the principles of the invention and the exemplication of embodiments thereof, that' a stabilization or linearization of amplifiers with inverse feedback can be accomplished so that there will be no tendency to self-oscillation or any other type of instability. In-brief, the basic idea underlying the invention may be stated to l involve the creation of a differential wavev by.l

combining in inverse Apha'se waves proportional to the input and output waves of an electrical or other translation system including an amplifier as is customary'in negative or inverse feedback amplifiers. Minute portions of the total energy content of the thus obtained differential `wave are selected sequentiallyat a' rate which is high compared with the highest component frequency of the waves being translated and the selected energy portionsare utilized to produce control energy applied to the input 'of the amplifier or translation system, said control energy varying as a function of the instantaneous magnitudes of the energy portions selected from said differential wave. According to one embodiment, Fig. 3, the sequentially produced control energy which will include a component substantially instantly compensating any distortion which may have arisen within the amplifier yor translation system, may constitute the only input for, the amplifier. According to another embodiment. Fig. 7, this control energy may be applied to the amplifier input in addition tothe input wave energy to be amplified or translated. If the conditions according to Equations 9 and 15 are complied with, the output of the system. as foi-- lows from Equation 10, will be entirely independent of the characteristics of the mainwave path or amplier A and will be dependent only on the characteristics of the feedback path B which may beA easily designed to have a substantially'linear input-output relation. i 1

What I claim is:

l. A wave translation system comprising aprincipal wave path including an amplifier, said wave path having imperfect transmissioncharacteristics. a lfeedback path-having substantially perfect transmission characteristics `for feeding back output energy in inverse phase upon the input of said principal wave path to producedifferential wave energy, means i'or sequentially We' y 7 selecting at a frequency which ls-high compared with the highest component frequency of the wave energy to be translated minute portions of vsaid differential wave energy, means for converting the selectedenergy portions into control energy having an amplitude undergoing successive variations by amounts proportionate lto the magnitudes of the selected energy portions, and means for impressing said control energy upon the input of said wave path.

2. A wave translation system comprising a principal wave path including an amplifier, said wave path having imperfect transmission characteristics, a feedback path having substantially perfect transmission characteristics'for feedingL back output energyin inverse phase upon the input of said principal'wave path to produce differential wave energy, means for sequentially selecting at a frequency which iss'high compared with the highest component frequency of the wave energy to be translated minute portions of said differential wave energy, means for converting the selected energy portions into'control energy having an amplitude varying proportionately ,tov the magnitudes of the selected energy portions, the time interval between two selecting periods being .larger than the sum of the transmission periods of said wave energy through said principal., and feedback paths, andv means for impressing said control energy` upon the input of said wave path.

3. A wave translation system comprising a principal wave path including an amplier, said wave path having imperfect transmission characteristics, a feedback path having substantially perfect transmission characteristics for feedingI back output energy in inverse phase upon the input of said principal wave path to produce difi ferential wave energy, n'ieansl for sequentiallyselecting at a frequency which is high compared with the highest component frequency of the wave energy to be translated minute portions vof said differential wave energy, means for converting the selected energy portions into control energy having an amplitude varying proportionately to the magnitudes of the'selected energy periods being larger than the sum of the transthe sole input for said wave path.`

4. A wave translation system comprising a principal wave path including an amplifier,l said wave path having imperfect transmission characteristics, means for impressing wave energy to be translated upon the input of saidwave path,

' a further wave path connected to the input of said. principal path, a .feedback path having substantially perfect transmission characteristicsA portions-the time interval between two selecting f for feeding back energy in inverse phase from the output of said principal path upon the output 4of said further wave path to produce differential wave energy. means for sequentially selecting at a frequency' which'4 is high compared with the highest component frequency of the wave energy being translated minute portions of said differential wave energy, means for converting the selected energy portions into control energy having an amplitude varying proportionately to the magnitudes of the selected energy portions, ,the

time interval between thetwo selecting being. larger than the of the transmission periodsof said wave energy through said prin,

cipal and feedback paths, and means for impressing said control energy upon the input of said principal wave path.

5. A wave translation system vcomprising a principal Wave path including an amplifier, said wave path having imperfect transmission characteristics, a feedback path having substantially perfect transmission characteristics for feeding back output wave energy in inverse phase upon the input of said wave path to produce/differential wave energy, means for sequentially selecting at a frequency which is high compared with the highest component frequency of the Wave energy to be translated minute portions of said differential wave energy, integrating means for converting the selected energy portions into control energy having an amplitude varying proportionately to the magnitudes of the selected energy portions, and means for impressing said control energy upon the input of said principal wave path.

wave path having imperfect transmission characteristics, a feedback path having substantially perfect transmission characteristics for feeding back output wave 'energy in inverse phaseupon the input of said wave path to produce correspending differential electric Wave energy, means for sequentially selecting at a frequency which is high compared with the highest component frequency of the Wave energy to be translated minute portions of said differential wave energy, an electrical condenser, means for successively charging said condenser by the selected energy portions, and further means for impressing the voltage developed by said condenser upon the input of said amplifier.

7. A wave translation system comprising a principal wave path including an amplifier, said wave path having imperfect transmission characteristics, a feedback path having substantially perfect transmission characteristics for feeding back output-wave energy in inverse phase upon the input of said wave path to produce corresponding differential electric wave energy, switching means for sequentially selecting at a frequency which is high compared with the highest component frequency of the Wave energy to be translated minute portions of said differential wave energy, an electrical condenser, means for successively charging said condenser by the selected energy portions, and further means for impressing the voltage developed by said condenser upon the input of said amplifier.

8. A wave translation system comprising a principal wave path including an amplifier,l said wave path having imperfect transmission characteristics for feeding back output Wave energy in inverse phase upon the input of said wave path to produce corresponding differential electric wave energy, electronic switching means for sequentially selecting at a frequency which is high compared with the highest component frequency of the wave energy being translated minute portions of said differential wave energy, a network comprising series resistance means and a parallel condenser, means for impressing the selected energy portions upon said network, and further means for impressing the voltage developed by said condenser upony` the input of said amplifier.

9. A wave translation system comprising a principal wave path including an amplifier, said wave path having imperfect transmission characteristics, a feedback path having substantially.

perfect transmission characteristics for feeding back output energy in inverse phase upon the input of said wave path to produce differential wave energy, means for sequentially selecting at a frequency which is high compared with the highest component frequency of the wave energy being translated minute portions of said differential Wave energy, means for converting the selected energy portions into control energy forming an amplitude varying in proportion to the instantaneous magnitudes of the selected energy portions, and means for impressing said control energy upon the input of said Wave path, the product of the transmission constants of said principal wave path, said feedback path, and said correcting means being'substantially equal te unity.

10. A wave translation system comprising a principal wave path including an amplifier, said Wave path having imperfect transmission characteristics, a feedback path having substantially perfect transmission characteristics for feeding back output energy in inverse phase upon the input of said wave path to produce differential wave energy, means for sequentially selecting at a frequency which is high compared with the highest component frequency of the Wave energy to be translated minute portions of said differential energy, the interval between successive selecting periods being high compared with the sum of the transmission periods of said wave venergy through said principal and said feedback paths, means for converting the selected energy portions into control energy having an amplitude varying in proportion to the magnitudes of the selected energy portions, the product of the propagation constantsof said main wave path and said feedback paths being substantially constant, and means for impressing said control energy upon the input of said Wave path.

11. A wave translation system comprising a principalwave path including an amplifier, said wave path having imperfect transmission characteristics, a feedback path having perfect transmission characteristics for feeding back output energy in inverse phase upon the input of said wave path to produce differential wave energy, means for sequentially selecting at a frequency which is high compared with the highest component frequency of the wave energy to be translated, minute portions of said differential energy, the intervals between the selecting periods being high compared with the total transmission time of said wave energy through said principal and feedback paths, means for converting the selected energy portions into controlx energy having an lamplitude varying in proportion to the magnitudes of the selected energy portions, means for adjusting the gain of said amplifier, further means for adjustingthe transmission constant through said feedback paths, and common control means for both said adjusting means tooutput wave energy in inverse phase with nonamplified input energy to produce differential wave energy, sequentially selecting at a frequency which is high compared with the highest component frequency of the wave energy being translated minute portions of said differential- Wave energy,'integrating the selected energy portions to produce control energy having an amplitudevarying in proportion to the ymagnitudes of the selected energy portions, and utilizing said control energy at least in part as said input enersy.

13.\Inr the art of amplifying and translating wave energy, the steps of combining amplied wave energy in inverse phasewith non-amplined wave energy to produce differential wave energy, sequentia1ly`selecting at a frequency which is high compared with the highest component frequency of therwave energy being translated minute portions of said differential energy, the selecting periods being small compared with the intervals between them, integrating t e selected energy portions to produce control e gy having an amplitude varyingin proportion to the magnitudes of the selected energy portions, and utilizing said control energy at least in part as said input energy. v

14. A wave translation system comprising a l principal wave path including an amplier, a

5 feedback path for feeding back ampliiled output energy in inverse phase upon the input of said \.wave path to produce diierential wave energy,

means for sequentially selecting minute energy.

{ portions oi said diierential wave energy, inte- 10 grating means for converting the selected energy portions into control energy having an amplitude varying in proportion to the magnitudes of the selected energy portions, and means for impressing said control energy upon the input of 5 said wave path.

GUs'rAvr: Gomma.

Images