US2210503A - Wave translation system - Google Patents

Description

WAVE TRANSLATION SYSTEM Filed Oct. 25, 1938 5 Sheets-Sheet 1 FIG. 2 Q 12 III-- mvavrox? By R. C. SHAW ATTOQNEV Aug. 6, 1940. R, c. SHAW 2.210.503

WAVE TRANSLATION SYSTEM Filed Oct. 25, 1938 5 Sheets-Sheet? FIG. 6 m;

LOW FREQUENCY REGION 5.5 REVOLUTION! OF SPIRAL. THESE SPIRALS OCCUR INA REGION OF THE COMPLEX PLANE NOT CONCERNED WI TH OPERATING STAB/L TY.

HIGH FREQUENCY REGION INFINITE NUMBER OF REV- OLUTIONS 0F SPIRAL.

A6 CHARACTEflIST/C JWTH RES/STANCE FEEDBACK SCALES DISTURTED r0 ILLUSTRA TE THE OPERA T/ON 0F c THE TRANSMISSION LINE,

FIG. 7

N0 FEEDBACK A 47 m 4s 9 B Q 45 E 4a TkANSM/SS/ON LINE LENGTH: {FEE-054cm 9.4 05. (2A 42 B TRANSMISSION LINE LENGTH 511;) 4| FEEDBIACK'. 8H7 l A a '40 l l -aoo -4oo o *400 *600 FREQUENCY INCREMENT-KC.

INVENTOR C. SHAW AT TORNE Y Aug. 6, 1940.

R. c. SHAW WAVE TRANSLATION SYSTEM Filed Oct. 25, 1938 5 Sheets-Sheet 4 F l6. l2

FIG/2A RT R FIG. /3

INVENTOR HAW By R C S ATTORNEY Aug. 6, 1940. R. c. SHAW 2.210.503

WAVE TRANSLATION SYSTEM Filed Oct. 25, 1938 5 Sheets-Sheet 5 S 3 FIG. /4 t; E \l 2 2 h 3 o FREQUENCY mcnsuswr- KC.

- u 3 FIG. /5 E 2 -l k 3. a

-|ooo 6 +1060 FREQUENCY nvcnznavr- KC.

VOL T465 GAIN- 0B.

FEEDBACK: I0 08.

-|o'oo b FREQUENCY llvmEMENT-kc'.

INVENTOR R. C. SHA W Patented Aug. 6, 1940 UNlTED STATES j PATENT OFFICE v v a ,210,503 7 V I WAVE TRANSLATIONJSYSTEM RobertjCaShaw, .H'olrn delpN. 1., assignor toBell- Telephone Laboratories, Incorporated, New York, N. Y., a corporation of New York Ap p lication "October 25, 1938, Serial. No; 236,857

ecl im gf (Cl. 179-471) The present inventionirelates to amplifiers. :Amplifiers including anegative feedback connection for feeding backaportion of the amplifier output to the input of one of the stagesof the amplifier, in such magnitude and phase as :to cause a substantial. reductioniin the over-all gain of. the amplifier with resultant marked im- 7 leads or'the physical lengthof the feedback loop.

Of. the several possible sources of deleterious phase shift, two appear to be of major importance: First, that due to the effect of electron transit time in the tubes; this effect is inherent in electron discharge devices and; thereforadt cannot be eliminated entirely but must be compensatedfor; second, thatdue to thephysical length of the feedback loop. For 'examplegin :high power short wave amplifiers'the size of.

tubes and the spacing of component parts ofran amplifier to meet high voltage insulation require.-

ments may result in a physical length of the b5- path of a negative feedback amplifier which is a considerable fractionof a wave-length. Asimilar conditicn may exist at audio or carrier fife: quencies in repeater systems where, feedback is desired over two ormore'amplifiers separatedby several miles of interconnecting lines. In most cases, the deleterious phase shift isprincipally a lagging phase shift in the ic-path of the amplifier and must becorrectediif maximum performaims is demanded fro'mthe: amplifier or trans:

mission system. i w p The unwanted phase shift may be. corrected, it has been determined, by usingluniped constants in the feedback circuit to introduce a leading phase shift providing the desired resultant phase shift around the feedback loop; or, it may be corrected by using a transmission line in the feedback circuit, or betweenstages of-th'e amplifier, the length of. the line being suchthat the necessary phase shift is "introduc'edto pro vide the desired resultant phase shift around the feedback loop. Aside from its-use incompensating for unwanted phase shift in an ultrahigh frequency amplifier, the transmissionline may be used'in any feedbackamplifier to' introduce a; desired or preassigned delay in the feedback circuit.

. Objects ofthe invention are to apply the negative feedback principle to amplifiers of high frequencies withresults comparable to those 010- 5 tained at relatively lower frequencies, and to compensate effectively. for unwanted phase shifts arising from the use of such high frequencies.

.A feature of. the invention comprises compensatingfor unwanted phase shift in the 1/ or 10 forwardly transmitting portion or path of an amplifier .by usein the florbackwardly trans-- mitting portion or path of the amplifier of a circuit of either lumped or distributed constants the: unwanted phase shift.

A'more complete understanding of the invention will befobtained from the detailed description which follows hereinaftentaken ini conjunction with the appended drawings, wherein:

Fig-1' is-aschematic of a transmission system including a feedback amplifier having a transmission linein its feedback path; v Fig. 2 is a schematic of a system similar to that introducing a phase shift that compensatesfor l of, Fig. 1 except that the amplifier is of thepush- .25

pul type; I p 1 i Fig. Elisa schematic of a system-similar to that of Fig. 1 except that an open-wire transmissionline is used rather than the concentric conductor transmission line of Fig.- l;

Fig. 4 isa circuit drawing of an ultra-high frequency amplifier includinga transmission line in ltsfeedback path; v g

.FigQ Bshows a pfi characteristic for the amplifieronFig. f1 with a resistance feedback only;

Fig. .6 shows an approximate mcharacteristic for the amplifier of Fig. 4 with a transmission lineprovidinga number of cycles delay in the feedback circuit; I

i the amplifier of Fig. 4 for conditions of no feedback and for feedback provided by transmission lines of different lengths;

Fig. 8 is a schematic of an ultra high frequency amplifier including a fl-circuit' having lumped constants for correcting for unwanted phase. shift dueto the effect of electron transit time; Y

Figs. 9,10, 11 and 12 are schematics of ampllfiers similar'to'that of Fig. '8' but including 5- circuits having different configurations;

*Fig. 12A shows theequivalent-circuit for the p circuitofFigylzg g #Fig. -13-is' a'circuitdrawing of an amplifier sim- 'ilar' totl'iatof'Fig-fil but havinga lumped con- 55.

stant 18-circuit .in place of a transmission line;

Fig. 14 illustrates the effect of the phase angle of; tube transconductance on the gain-frequency characteristics of the amplifier of Fig. 13 in the case of a resistance feedback; and

Figs. 15 and 16 illustrate the symmetrical gainfrequency characteristic that can be obtained with the amplifier of Fig. 13 when the p-circuit is designed .to compensate for the-phaseangle of tube transconductance.

The ip-characteristic of a feedback amplifier operating atlow frequencies may be expressed as lbfi=lfiltl |0+ I where: 3] is the absolute value of as; 0 is the phase advance or lag due to circuit elements and which is variable with frequency; to is a constant phase, due, for example, to the vacuum tubes, invariable with frequency. In the simple three-stage, shunt-shunt feedback amplifier, =3 l80 degrees or 180 degrees. 'At some particular frequency, usually at mid-band, 6 is made equal to zero and the feedback voltage is exactly 180 degrees out of phase with the applied voltage, As the frequency'is varied the phase 0 will also vary and will be added to or subtracted from the constant'pha'se (p; The desired characteristic in the pfi-IOQP may be obtained by using the principles of H. W. Bode Patent 2,123,178, issued July 12, 1938.

Turning to amplifiers operating at higher frequencies a different problem exists. The phase (p which was termed constant at low frequencies now becomes dependent on frequencyand in many instances becomes anappreciable fraction of a cycle.

ticular amplifier, that is, if the time offtransit of the electrons from the cathodeto the anode becomes an appreciable portionof a cycle of the operating frequency; an effective delay is introduced into the amplifier tube trans'conductance. Thus the developed platefvoltage, with a resistive plate load, will be out of phase with the grid voltage by degrees minus m-degrees instead of the usual 180degrees, where-thelagging angle in the tube transconductance is (p'r degrees. In; the published work of F. B. LlewellyniiProceedings of the Institute of Radio Engineers, vol. 22, August 193 i) it is shown" that the transit angle of a triode may be expressed as follows:

trans tangl I. where:

k and 7c are constants x=cathode to anode spacingcm. I A=operating wave-length -cm.

o=anode voltage-volts f=frequencyc. p. s. ,1 a I If these factors are known Llewellyn shows that the transit time maybe. calculated-and turn the effective phase angle p'r 0f the transconductance determined. Referring tothe above equation, it may be noted-,thatthe transitangle a i th v ta e .m :W u: freq nc e variations due to voltage changes may be neg- 63''meg'acycles, (111,:400 v.) had an effective measured angle ga'r in its transconductance of about 15.3 degrees. Three stages would have about 46 degrees. This means that at midband frequency, in an amplifier using these three stages, the input grid voltage and the output plate voltage are out of phase by 180 degrees minus 46 degrees or about 134 degrees. If this output voltage is fed back directly to the grid of the first stage (with no change in phase) the desired condition for feedback at mid-band frequency' (input and output voltages 180 degrees out of phase) is not fulfilled. Both gain and phase margins against instability are reduced. The resultant effect is a reduction in the amount of feedback before singing occurs, bad dissymmetry in the gain-frequency characteristic, and a decrease in the width of the useful frequency band. v

The equation for this amplifier may be expressed as:

where I c], and [L are the same quantities as in the low frequency case, and

to the efifective phase angles of the tube trans.- conductances. e

may also be due to the fact that the apparatus is an appreciable fraction of a wave-length in size. It may be "noted that this ip-equation is identical with the low frequency condition except for the multiplication by a phase shift due to transit time.

Before considering methods of compensation for the angle it is desirable to know its value not only at midband frequency but also at the critical frequencies of gain and phase cross-over. The ideal compensating circuit would be one which corrected exactly for all frequencies. This is theoretically possible but inmost practical cases such perfect correction is not needed and simple. circuits providing approximate compensation will suffice. For convenience, we may divide the methods of correction into two classes: (1) Those dealing with amplifiers where the variation of the condition where the variations order of afew per cent; (2) those dealing with in For amplifiers in class (1) the design would be made in two steps. The phase angle 4 1 would be compensated for by methods discussed hereinafter. Then the amplifiercircuits would be designed as suggested in the Bode patent.

For amplifiers" in class .(2) the excess phase shift w would be compensated for by altering the design of the amplifier using Bodes methods.

Methods of compensation for which might be used in the first class will now be considered. For simplicity, the correction will be'made first for a single frequency. This may be done by increasing the delay inthe ,uB-IOOP so that the total phase lag will be a complete cycle or 360 degrees. This can be accomplishedby using a distributed constant network, such as a transmissionv line. The line introduces phase shift with negligible attenuation.

This is illustrated by Fig. 1. The transmission line 10 has a source H of signal waves connected across its input terminals, and a suitable load or terminating impedance I2 connected across its output terminals, a feedback amplifier I3 being connected in the transmission line. The amplifier, for example, for amplifying very high frequencies, of the order of megacycles, comprises a ,uor forward path that may be, for example, a three-stage amplifier, a'nda feedback connection I 4. Ordinarily, the feedback connection would include merely the transmission element 15, for instance, a resistance or-other impedance, labeled ,8. To compensate for the undesired phase shift in the epath, resulting from the effect of electron transit time in the tubes in the -path, thetransmission line or concentric cable 16 of length Zis introduced in the feedback connection and may be considered as part of the por backwardly transmitting path of the amplifier. The transmission line would be terminated, preferably, in a resistance equal to its surge impedance. The

length Z of line inserted should be equal to where A is the wave-length at which the correction is desired,

is the undesired phase shift in degrees and n is a positive integral number, 1, 2, 3, etc. The ,upequation may now be written as be only "If desired'jthe termination of the compensating line maylbe different from aresistance equal to its'1surgeimpedahce provided that the new termination'is taken into account, using well-known transmission line formulae, in the ip-equation in setting up the requirements for feedback. Under thiscoridition, the required length of transmission line and its value of surge impedance will depend on the characteristics of the terminating impedance. n this instance the transmission linein conjuriction with its terminating impedance might serve the dual functions of the phase shift and attenuation required in the ,B-section of the amplifier. 'Otherphysical characteristics of the line, such as "the material of which it is constructed, may' be altered toproduce desired results. f' 1 "Itis realized that if the unwanted phase shift I 1 s i' should be positive in a particular transmission system, the length of line required for compensation will be 1.

In the example described above the compensating line was shown connected in the p-path. It is possible to insert the line at any point in the ie-loop. For example, it may be in the ,IL-- or forwardly transmitting portion or path, and might be inserted between stages of a multistage amplil )\(.5m where m. is an odd positive integer, 1, 3, 5, etc., \=wave-length at which correction is desired, and

is the total undesired phase shift in degrees. Fig. 2 is representative of a push-pull amplifier, for example, of a single stage with the feedback connection from the output of one tube to the input of the other tube. In each feedback connection, the length of the transmission line need instead of Fig. 4 shows the application of the invention to a multistage negative feedback amplifier, designed to operate at 63 megacycles and embodying an open-wire transmission line.

'The amplifier comprises a three-stage pushpull amplifier, each stage comprising a twin pentode I8 of the typedisclosed in A. L. Samuel Patent 2,062,334, issued December 1, 1936. The interstage networks I9, 20 are of simple, anti-resonant type. Control grid, plate and screen grid potentials are supplied from sources Eg, Eb and S50, respectively. Heating current for thecathodes, which, of course, could be of the indirectly heated type, is supplied through points A, A. The source 2| of high frequency signal to be amplified ma be inductively coupled through the windings 22, 23 to the input circuit of the first stage of the amplifier, and the load 24 to which the amplified signal is being transmitted may be inductively coupled through the windings 25, 26 to the output circuit of the last stage of the amplifier. A feedback connection comprising an open-wire transmission line 21 connects'the output of the amplifier with the input of the amplifier, the line 21 being terminated at the input of the amplifier in a resistance R equal to the surge impedance Z of the line, the equal resistances R2 each being much larger than R. The input impedance of the line serves asv the shunt element of the fi-circuit potentiometer and the equal resistances R1 are connected between each side of the line and the output plate circuit to serve as series elements of the c-circuit potentiometer. The resistances R2 also serve as series elements of this potentiometer.

The operation of the transmission line in the c-circuit of this ultra-high frequency amplifier may be described as follows. The feedback signal, originating in the plates of the last stage, is transmitted through the line to the input circuit of the first stage. The line introduces phase shift with negligible attenuation. Thus, the effect ofthe line is to rotate the phase of the fe'd back voltage by an amount depending on the length of line. If the length of line is one-half wave-length, the phase shift will be 180 degrees at mid-band frequency. For frequencies well separated from mid-band, however, different phase shifts will occur. If the gain remained large, these-might not be desirable. In practice, however, such latter phase shifts occur in a region of the frequency spectrum which has little effect on the stability. Thus, with certain limita tions, any integral number of half wave-lengths of line may be used.

Returning the feedback through a transmission line offers a number of advantages. The output and input circuits of the amplifier may be physically separated. This eliminates the strict requirement heretofore imposed in feedback amplifier design that the path of the voltage fed back be not only as short as possible but be negligibly short. At ultra-high frequencies, as in high power short wave systems, of course, this is of major importance. Furthermore, it is possible at one frequency to introduce in the fl-circuit of an amplifier any phase angle whatsoever by the simple expedient of adjusting the length of line to the required number of electrical degrees. The use of a transmission line, also, would permit feedback .around antennas, thus including feed lines and antennas within the ie-loop so that they can be operated on by the feedback to improve or change their characteristics.

The performance of an amplifier with a "transmission, line in its p-circuit is best analyzed by means of its ip-characteristic. Consider the three-stage amplifier shown in Fig. 4. With a pure resistance feedback and no transmission line therein the lacharacteristic may be expressed by the equation;

Plotted for a particular. set-up of the amplifier, the curve of Fig. 5 is obtained. If a transmission line of length I is inserted in the B-circuit, the general equation for pfiTL=l is where A is the wave-length at the particular frequency for which MBTLzZ is being computed. The attenuation of the line is assumed negligible. The ,uBTL=Z-Cha1'a0telliSfi0 will be a system of spirals: the low frequency region extending from zero to the neighborhood of the useful band will have a finite number of revolutions equal to the number of mid-band wave-lengths in the transmission line; the high frequency region,-extending from the useful band to finite frequency, will have an infinite number of revolutions. Such a characteristic for an amplifier having a transmission line of 5.5 wave-lengths is shown in Fig. 6. For this line the phase shift in the fl-circuit at mid-band frequency is 1980 degrees. In most practical cases, the stability of the amplifier with the line will not be greatly affected from the condition existing without the. line. This results from the fact that the inner revolutions of the fr-characteristic, as plottedjin the complex plane, will occur in a region which is not significant in the consideration of operating stability: namely, well within a radius of unity from (0,0). Difficulties may be expected, however, when the differential phase shift introduced by the line in the frequency range 'of interest, becomes large. This phase shift may be readily calculated. Let the length of line =1, and the time delay =7.

where c=velocity of propagation.

The differential phase shift ibL-m from the midband frequency fm to some other frequency f1 will be V i i-m Where and "Y (fl f o) where )vy is the wave-length corresponding to the differential frequency 'y.

If the tolerable phase shift at some critical frequency is known, the maximum permissible length of line can be determined.v For example, if the limiting phase shift is radians at a frequency removed from mid-band 3 10 1:5 meters.

Kit issassunedthatina-particular amplifier, the

mid-band frequency is 150 megacycles, i. e., Am=2 meters; the maximumllength of transmission line would be about 3 wave-lengths.

The actual performance of an amplifier with transmission line feedback was investigated in an'amplifier in: accordance with-Fig. 4. First, a. 0.5). line was used and itsv length changedto curve of the amplifier be symmetric about the mid-band frequency. The use of a line an odd number of half wave-lengths in length requires one transposition to obtainthe correct phase relations. This was the condition in each test.

In the first case, it was found necessaryto reduce tlielength of the line from or 238 centimeters to 184 centimeters in order to obtain'th'e symmetric gain-frequency curve A of Fig; 7. This corresponds to a change of 40 electrical degrees and compensates for the phase shift due to the effect of transittime in the tubes. In making this measurement, all capacitive effects across the line were neglected, and the 1 amplifier-was operated withthe plate'and screen grid voltages at half normal value.

There was no' evidence of instability.

In the second case, the plate voltage was inl creased on two stages of the amplifier by a factor of two. A symmetric gain-frequency characteristic was obtained by increasing the length of the transmission line from 184 centimeters to 19'? centimeters, corresponding to a change of 9 electrical degrees, or to a decrease in the transit angle in each of the two stages of 4.5 degrees. According to theory, there is to be expected as a first approximation a decrease per stage of or 5' degrees. Measurement, therefore, was in good agreement with. theory.

In .the third case, 5.5x of line replaced the 0.5x section. To obtain the proper compensationfor the phase angle of the transconductance, the line was reduced in length by about 65 centimeters. The gain-frequency characteristic of the amplifier with this long line (5.5X63mc. =26.18. meters) in its feedback circuit is shown. by curve B of Fig. 7. Curve C is a gain-frequency characteristic for the amplifier of Fig. 4 with .no feedback present.

From what has. been detailed hereinabove, it is apparentth'at itis feasible to. insert a transmission line of one or more ,half wave-lengths in thee-circuit of a feedback amplifier, and that by adjustmentof its length compensation may be provided for the phase angle of the. transconductances of the tubes. In high power short waveamplifiers, a

line may, provide a practical feedback. path with: out rnateri'ally reducing the .amount. of feedback possible. Furthermore, the line may be used to measure. transit angles and changes in transit angles, and to introduce any desired number .of cycles delay in a feedback circuit.

It is known in the art that the over-all envelope delay, as distinguished from the delay around the loopgof a feedback amplifierusing a large amount of feedback, is the negative of the envelope delay in the B-circuit. Thus, an envelope delay in the p-circuit would actually result in an advance of the envelope. It seems possible, therefore, that a feedback amplifier with a transmission line in its ,e-circuit might be used to provide a desired envelope advance, e. g., an envelope feedback loop with a radio frequency feedback loop inside it,- thearadio' frequency loop containing a radio frequency transmission line in its 5- circuit.

In manyinstances it willbe more convenient touse circuits containing lumped constants for effecting compensation or correction for the deleterious phase, shift in a transmission system by introducing a leading phase shift in the ,cc-path. The circuit may perform the function, also, of providing the desired attenuation in the s-path. That is, the circuit'may serve both asthe B-path potentiometer and the phase correcting means.

Fig. 8 is a schematic of a three-stage high frequency amplifier 30 having a feedback or B-path 3|. The amplifier is shown in a transmission line orpath 32, a source 33of high frequency signal being connected to the terminals of one end of the line, and a load 34 into which the amplifier works being connected to the terminals of the other end of the transmission line. The s-path 3] comprises a series resistance Rs and a shunt impedance Z comprising a resistance R and an inductance L. in parallel. R5 and the impedance looking toward the input grid from the shunt circuit R'| |L should be high with respect to Z and the impedance of the B-path should be large with, respect to the plate impedance in theoutput of the amplifier. This is the condition usually metin practice. If such is not the case, the simplified equation. used below cannot be employed. The familiar general circuitequations may thenbe used which take all elements into consideration. From these equations the magnitude and. phase of the fed-back voltage a relative to the applied voltage ea may be expressed in terms of the circuit elements in the e-network. From this, the correct values may be assigned to the elements in the c-network to provide thedesired-amplitude and phase of the fed-back voltage at. c For the condition usually met in practice, the voltage, e impressed on the input grid is equal to the signal voltage es plus 1 the fed-back voltage a A current --i g c R.+z iiows in the feedbackcircuit, and with The/.ratioof..theimpedances of the shunt section Z to the series section Ra -i5 selected to givethe desired attenuation in these-path; the phase angle of the shunt section is chosen so that it is equal to Z g g} the total unwanted phase shift resulting from the effect of transit time. If, in the particular case,

=45 degrees then tan" =45 degrees and R=L. This circuit may be arranged to compensate exactly for Q at mid-band frequency, introducing a leading phase shift to compensate for the lagging phase shift due to the effect of transit time.

Fig. 9 shows another circuit3 la for the p-path. The series section comprises a capacitance Cs and the shunt section Z comprises a resistance R and a capacitance C in series. If

. P5 is the phase angle of the tube transconductances to be corrected for, the phase angle of Z should be chosen such that i l V tan" =9[ degreesp, In Fig. 10, the circuit 3lb for the [Ex-path is shown as comprising a capacitance Cs for its series section and its shunt section a capacitance C and a resistance R in parallel. Here 1 1 R fz audits phase angle should be.

tanwC'R=90 With the arrangements of Figs.'8, 9 and 10, compensation may be made for angles up to about degrees. It might be pointed out that the magnitudes of Z and the phase correcting angles may not be symmetrically displaced about the mid-band frequency, as the frequency is varied, but in most practical cases this lack of symmetry will not be a source oftrouble.

In Fig. 11, there is shown a lumped constants circuit 3lc for providing corrective or compensating phase shifts of from. 90 degrees to degrees; In this case, the 5-series section is a capacitance CS and the shunt section comprises an inductance L and a resistance R in' series,

Z= /R +X the phase angle of Z being tan" the operation ofthe circuit being similar t that of Fig. 8. C1, C1 are by-pass condensers, B1 and B2 areplate and filament heating battery, re-' spectively.

The circuit of Fig. 1 2 could be employed, also,

to compensate for variations in due to voltage changes in the ,u-path of the amplifier. The characteristic of the tube T would be such that as the voltage fed back to its plate varies in amplitude, the phase of the effective interval impedance of the tube T varies to compensate for the variation in Although the tube T is shown as a diode, other types of tubes may be used.

Fig. 13 is a circuit drawing of a three-stage amplifier designed to amplify a signal frequency of 63 magacyoles and to provide a feedback of 10 decibels. Its similarity to that of Fig. 4 is evident, like parts bearing corresponding identifying characters, but the transmission line feedback is not included. It has been determined that if no compensation is provided for the phase angle of the tube transconductances, there is bad dissymmetry in the gain-frequency characteristic, a reduction in the permissible amount of feedback before singing occurs, and. a

decrease in the width of the useful frequency band. This is brought out clearly when the p-circuit of the amplifier of Fig. 13 consists of a pure resistance'in its series section and a pure resistance to ground in its shunt section. Fig. 14 shows the gain-frequency characteristic for such an amplifier for a condition of no feedback, or negative feedback at mid-band of about 4 decibels, and of negative feedback at mid-- band of about 10 decibels; the feedback extends over a range from about 62 to 64 megacycles.

Similar results would be obtained if the re-v sistances of the [3-Circl1i'b were replaced by pure capacitances. When the B-circuit comprises, however, one of the transit time phase angle compensating circuits of Figs. 8 to 12A, a symmetrical gain-frequency characteristic is obtainable. Using the p-circuit of Fig. 9, the amplifier of- Fig. 13 had the gain-frequency characteristic of curve A, Fig. 15, taken for a negative feedback of 4 decibelswith the phase angle of the tube transconductances exactly compensated at the mid-band frequency of 63 megacycles. In the particular amplifier, the phase lag due to the effect of transit time was about 46 degrees. The combination of resistance and capacity in the B-shunt section was designed to have a phase angle of 44 degrees, i. e., 90' degrees minus 46 degrees, the voltage developed across the shunt section lagging the current through the shunt section by 44 degrees. When the parallel resistance and condenser of Fig. 9 werereplacedby a resistance and the parallel arrangement introduced in the series section in place of the condenser, the symmetrical gainfrequency characteristics of Fig. 16 were ob-v tained for conditions of mid-band feedback of 4.4 decibels and of 10 decibels.

In connection with Fig. 8,,it was stated that the fl-circuit arrangement of I that figure could be proportioned so as tocompensate exactly for Pr I at mid-band frequency. This may not be the condition, however, at the critical frequencies of gain or phase, cross-over, It ispossible, how-- ever, to select circuits so that the phase margin. may be improved. This may be doneby using wanted angle. This is illustrated in the table below which shows the variation in g2} (arbitrarily assuming g} af across the critical band of a 63-megacycle amplifier such as that of Fig. 13 using the compensating circuits of Figs. 9 and 10.

From this table, which shows the effect of two different compensating circuits whose phase angles vary in opposite directions with frequency,

3 it is evident that improvement is obtained in a particular case by selection of the most appropriate circuit. If more complicated feedback circuits are employed, exact compensation over the entire critical frequency band may be obtained.

Although this invention has been disclosed with reference to various specific embodiments, it is to be understood as not limited thereto but by the scope of the appended claims only.

What is claimed is:

1. An amplifier comprising a or forward path having a phase shift due to the effect of electron transit time, and a cor backward path including means to compensate for said phase shift.

2. An amplifier as claimed in claim 1 in which the phase shift is a lagging phase shift, and said means comprises a lumped constants circuit introducing a leading phase shift correcting for said lagging phase shift.

3. An amplifier as claimed in claim 1 in which said means comprises a transmission line of length where Z=length in meters, A is the wave-length in meters of the mid-band frequency of the frequencies amplified, n is a positive integer, 1, 2, 3, etc, and p is the total phase angle for which compensation is to be provided.

4. In a high frequency amplifying system, an electron discharge device having an input electrode and an output electrode, a source of high frequency energy connected to said input electrode, the frequency being so high that there is a phase shift additional to the normal phase shift between the input and the output voltage of said device, and a negative feedback circuit compensating for said additional phase shift.

5. In a high frequency amplifying system, an electron discharge device in which the energy to be amplified is of such high frequency that the phase shift between the input and output voltage of said device is 180 degrees minus g0 degrees, where c degrees is a phase shift due to the effect of electron transit time, and a feedback circuit including means compensating for said phase shift. l

6. An amplifying system comprising an amplifier including means for introducing a phase shift of 180 additional to its overall phase shift and a feedback connection between the output and input circuits of said amplifier, said connecnection including a transmission line of length where l=length in meters, t=the amplifiers midband frequency, m is an odd integer, and (p is a phase shift in the amplifier due to the effect of electron transit time.

7. An amplifying system comprising an amplifier including means for introducing a phase shift of 11' radians additional to the normal phase shift between the input and output of said amplifier, and a feedback connection for said system including a transmission line of length Z=.5m A, where \=the mid-band frequency of the signal to be amplified, m is an odd integer, and Z is in meters.

8. A transmission system comprising a ,uor forwardly transmitting path and a 3- or backwardly transmitting path, said t-path having a phase shift therein due to the effect of electron transit time and said p-path containing means compensating for said phase shift, said means including an electron discharge device.

9. An amplifier having a or forwardly transmitting path, and a sor backwardy transmitting path, said n-path comprising an electron discharge device, the phase shift between the input and output of said device including a phase shift because of the effect of electron transit time at the frequency being amplified, and means in the B-path to compensate for the transit time phase shift, said means comprising a series impedance and a shunt impedance, said shunt impedance comprising an electron discharge device having a cathode and an anode, and a choke coil connected between said cathode and anode.

ROBERT C. SHAW.

Images