US2480163A

US2480163A - Negative feedback amplifier

Info

Publication number: US2480163A
Application number: US588269A
Authority: US
Inventors: Romander Hugo
Original assignee: Standard Telephone and Cables PLC
Current assignee: STC PLC; Federal Telephone and Radio Corp
Priority date: 1945-04-14
Filing date: 1945-04-14
Publication date: 1949-08-30
Anticipated expiration: 1966-08-30
Also published as: GB610105A; FR925551A; BE464915A

Description

Aw. 3m,

H. ROMANDER 2,480,163 y NEGATIVE FEEDBACK AMPLIFIER Filed April 14, 1945 3 Sheets-Sheet l I] V14 5 our/Our s 2 W M14/N AMPM/75? l/ 4 I j@ 6 y 3 IZ IN VEN TOR. H060 /Po/w/f/voE/P BY am v1.30, W49. H. ROMANDER n 2,430,163

NEGAT IVE FEEDBACK AMPLIFIER Filed April 14, 1945 s sheets-sheet 2 IN VEN TOR. HUGO ROM/INDE? Mgg 309 m9.

Filed April 14, 1945 H. ROMANDER NEGATIVE FEEDBACK -AMPLIFIER 3 Sheets-Sheet 3 IN V EN TOR.

HUGO ROMA/mf# AGENT Patented Aug. 30, 1949 NEGATIVE FEEDBACK AMPLIFIER Hugo Romancier, North Caldwell, N. J., assignor to Federal Telephone and Radio Corporation, New York, N. Y., a corporation of Delaware Application April 14, 1945, Serial No. 588,269

6 Claims.

This invention relates to wave amplifying systems and in particular to amplifiers employing negative feedback. Specifically the invention relates to a method of preventing self-oscillation in amplifiers of this type.

It is well known that in feedback amplifiers the phase shift which waves of different frequencies undergo during one complete excursion of their travel around the so-called a path varies for the different frequencies. It is usually desirable to so design an amplifier that al1 waves traversing the a path over a given range of frequencies undergo a phase shift of substantially 180 or as close thereto as possible. However, even though these conditions are fulfilled, there are always some frequencies outside of the given range for which the phase shift becomes zero, 360 or a multiple thereof, and if the amplifier gain at these frequencies is greater than unity the amplier may sing or self-oscillate.

It has been suggested that amplifier self-os-cillation may be prevented by including within the amplier circuits, and usually in the feedback path, a network so designed that the waves which tend to produce self-oscillation are highly attenuated thus reducing the gain at these frequencies. However, with my invention self-oscillation at all frequencies is prevented in accordance with a new principle wherein there is produced in addition to the -conventional feedback voltage, an additional voltage having such magnitude and phase that the resultant feedback voltage is always displaced 180" with respect to the input voltage.v It is immaterial whether the waves which traverse the amplifier are originally impressed thereon from an outside source or are the result of modulation or distortion components developed within the amplifier.

One of the primary purposes of employing negative feedback is to reduce distortion components developed within the amplifier. With the circuit arrangements of my invention the distortion components are most effectively reduced in the frequency range where the deviation from the desired 180 phase shift for one complete excursion around the MS path is relatively small, that is, of the order of 30 or less. However, since it is at these frequencies that distortion due to hum voltages occur, my circuit arrangement is very effective in reducing this type of distortion. Negative feedback may be applied almost without limit with the result that the h-um voltages are practically eliminated and still the amplifier will not oscillate at any frequency.

A primary object of my invention is to provide a stable negative feedback amplifier circuit arrangement which permits high gain over anextremely wide frequency range.

Other -objects are:

To provide an improved negative feedback amplier in which the phase difference between an applied input voltage and the net input voltage is substantially zero.

To provide an improved negative feedback amplifier in which distortion components are suppressed over a limited frequency range while the amplifier remains non-oscillatory over an extremely wide frequency range.

To provide a negative feedback amplifier having an overall gain which is proportional only to the forward amplification or It.

To provide a negative feedback amplier in which the ratio of the net or effective input voltage with feedback to the applied voltage is constant.

Other objects and features of my invention will become apparent from the following description, the appended claims and the drawings in which Fig. 1 is a schematic diagram illustrating a preferred embodiment of my invention;

Figs. 2, 3, 4 and 6 are vector diagrams which illustrate the principles of operation of my invention;

Fig. 5 is a polar diagram showin-g the ,variation of m8 with frequency; and

Fig. '7 is a schematic diagram illustrating a modification of Fig. l;

Referring to Fig. 1 the reference character I represents a source of voltage which is to be amplii-led. This source may be of a single frequency or of a plurality of frequencies such as would occur in the output of a voice frequency amplifier. Although the source is illustrated as an alternator it could also be the secondary winding of an input transformer or the output circuit of a vacuum tube amplifier. The reference character 2 represents the impedance of the source I and may be, for example, a separate resistance as shown or the plate impedance of an amplifier. The resistance 2 is connected to a junction point 3 and to the grid 4 of an amplifying tube 5. The amplifying tube 5 is illustrated as a triode for simplicity although tetrodes or pentodes could be employed without departing from the principles of my invention. The cathode 6 of the tube 5 is connected to ground through a biasing resistor 1. The anode 8 is connected to a main amplifier 9 illustrated in block form. The internal construction of the amplifier is immaterial in so far as my invention is concerned, It may b e a` single or a multistage amplifier; if the latter, the various stages may be transformer coupled, resistance-capacity coupled or by any other forms of coupling. Local negative feedback circuits may exist within the amplifier 9 and the phase shift of the amplifier at various frequencies may have any values whatsoever a1- though it would be preferable to employ the best practice and make the phase shift other than the 180 required for negative feedback as small as economically desirable. The amplifying tube is to be considered as part of the amplifier 9 but it is shown separately for purposes of description.

It is understood, of course, that for an unbalanced negative feedback amplifier such asis illustrated in Fig. 1 and where therev is substantially zero phase shift in the feedback path, a phase reversal of the voltages for the frequencies within the operating range must occur in order to reduce distortion in accordance with the principles of negative feedback. All of the phase reversal need not take place inthe forward or n path; some of it may be in the feedback or path.

In amplifiers of the push-pull type the phase reversal may be obtained by crisscrossing the feedback connections from the output to the input circuits in accordance with known practice. The principles of myV invention apply equally well to amplifiers of this type.

The necessary power supplies for operation of the amplifier are not shown` but are understood to be within the amplifier 9.

The output impedance of the amplifier 9 is illustrated as a resistor I0. A lead II connects the output of the amplifier to a load circuit, not shown. From a variable tapv I2 on the resistor Ill a negative feedback connection comprising a blocking capacitor I3 and a resistor I4 connects with the junction point 3. As described so far the circuit arrangement constitutes a typical form of feedback amplifier such as may be found in the prior art. The gain of the amplifier at various frequencies may besuch that the amplifier will oscillate at some frequency, perhaps not in the range for which the amplifier is designed but at some higher or supersonic frequency at which the phase rotation through the amplifier and feedback path is 0, 360' or a multiple thereof as above explained.

In accordance with my invention I provide a circuit arrangement, also connected to the junction point 3', which will produce a voltage having the proper magnitude and phase which, when combined with the feedback voltage of the main amplifier 9, will always produce a resultant feedback voltage having a 180 phase displacement with respect to the inputvoltage. The circuit arrangement is so designed that it will function in the desired manner at any frequency whether this frequency be that of the input voltage or of a distortion voltage developed within the amplifier 9. This circuit arrangementA will now be described.

From the junction point 3 a conductor I5 leads to the input grid I6 of amplifying tube I1. The cathode I8 of this tube is connected to ground through a resistor I9'. TheV anode: 20 connects to what I have termed a correlativo amplifier 2|, also shown in. block form. The tube I1 is to be considered as part of thecorrel'ative amplier but is separately shown forv purposes of description. The output of the correlative amplifier is connected to an output impedance shown as a resistor 22. A power supply, notl shown, is to be considered as connected to the tubes Within the 4 correlativo amplifier. If desired the power supply for both amplifiers 9 and 2I may be common.

The characteristic feature of the correlativo amplifier 2i is that it should have substantially identical gain and phase shift characteristics with respect to frequency as does amplifier 9.

The feedback connection comprising capacitor I3 and resistor I4 merely impresses a predetermined fraction of the output voltage developed across resistor I@ onto the input of the main amplifier. This fraction is a scalar quantity, that is, no phase shift is involved. For this reason a similar feedback circuit is not shown in connection with the correlativo amplifier 2|, any difference in voltage developed on the output resistor 22 due to this omission being compensated for by a suitable adjustment of the contact 23 which selects the required output voltage. On the other hand, if the feedback connection associated with the main amplifier is such that a phase shift occurs in this connection, a corresponding circuit producing the same phase shift should be associated with the correlative amplifier.

There may be cases wherein the construction of the correlatlve amplier may be simplified by, for example, the omission of a stage of amplification which involves nothing more than a phase shift of This phase shift could be compensated for by reversing the connections of a transformer within the amplifier or by other suitable means.

In practice amplifier 9 would usually comprise tubes capable of producing considerably higher output than the tubes in amplifier 2|. The variance in power output between amplifiers 9 and 2i is permissible since amplifier ZI merely supplies a voltage of desired magnitude and phase, and the value of this voltage, as determined by the point of contact 23 of conductor 24 on resistor 22, may be the total output voltage of the amplifier, whereas in conventional feedback practice the feedback connection at contact I2 on resistor I0 in the output of amplifier 9 is usually a very small percentage of the output Voltage. The essential point is that the gain and phase shift characteristics with respect to frequency, from the junction point 3 to the output of amplifiers 9 and 2I respectively, be substantially identical.

There is one essential difference between ampliers 9 and 2l and this pertains to their manner of operation rather than to their structure. The amplifier 9, being in fact a power amplifier is usually operated at high efciency with the result that the amplifying tubes of said amplifier operate over a large portion of their characteristics, and if their characteristics are not linear, distortion and modulation products are produced. On the other hand it is not necessary that amplifier 2l operate at high efficiency. The tubes of this amplifier should operate only over the linear portion of their characteristics, it being essential that no appreciable amplitude distortion occurs within the amplifier.

Conductor I5 also connects junction point 3 with the grid 25 of tube 29, this latter tube consistuting together with tube 2l a phase mixer 28. The grid 29 of tube 2l is connected with contact 23 of resistor 22 by the conductor 2li. The cathodes 30 and 3l of mixer tubes 26 and 21 are connected to ground through resistors 32 and 33 respectively, these resistors being shunted by capacitors 34 and 35. The resistors 32 and 33 develop the desired negative biases for the grids of their respective tubes.

Resistors

36 and 31, connected to the anodes 38 and 39 of mixer tubes 26 and 21 respectively, are for isolating the output voltages of these tubes so that they will not interact one with the other but will combine to form a resultant voltage having a value which is the vector sum of the voltages developed by tubes 26 and 21. This resultant voltage which appears at junction point 4U is amplified by the two-stage amplifier 4i. Resistance coupling is employed in this amplifier in order that a rninimum of phase rotation of voltages will occur therethrough. Two stages are employed in order that the phase of the voltages developed in the output of amplifier 4| will be the same as those applied to the input; The output voltage of amplifier 4I is developed across the resistor 42 in the plate circuit of the second stage. From a variable contact 43 a conductor 44 connects to the junction point 3 through' blocking capacitor 45 and resistor 46. The impedance of conductor I5 is assumed to be negligible.

In the above discussion -reference has been made to resistors 2, I4 and 46. These resistors permit the use of a single grid on which the lcombined voltages of three separate sources may be impressed. It is assumed that the resistances are all equal and have values sufficiently great to permit the generation of voltages at each source uninfluenced by the voltages generated at the other two sources. In order to simplify the explanation of my invention I have assumed that the resistances 2, i4 and 46 are all equal so that the three voltage sources each have equal opportunities to iniiuence the net voltage on grid 4, although an increase in any of the resistors can be compensated for by a proportionate increase in the respective source voltage. Moreover, such phase rotation as may exist in practice due to tube and stray capacitance at point 3 to ground is disregarded since this effect has been found negligible over the frequency range in which self-oscillation due to feedback around the main amplifier is possible.

The operation of the circuit of my invention will now be explained by making reference to the vector diagram shown in Fig. 3. In the following discussion of feedback amplifiers, and in particular such amplifiers utilizing my invention, certain mathematical notations and expressions will be used and are here defined. Reference characters found in the definitions refer to Fig. 1.

Notation or Expression Denition The voltage gain of the main amplifier without negative feedback. This is a vector quantity, having both amplitude and phase angle.

The factor by which the voltage output of an amplifier employing feedback must be multiplied to obtain the voltage which is impressed back onto the input of said amplifier. This, also, is a vector quantity.

The phase angle of the product p3.

A reference frequency, generally Within the useful frequency range of the amplifier, at which qS is 180.

A rfaegice frequency at which 15 is zero or a multiple In Fig. 2 I have represented by vectors the system of voltages which add vectorially to produce a net voltage V at the grid 4 of amplifying tube 5. Thus, at any given frequency, there will exist simultaneously the applied or actuating voltage V1, the feedback voltage V2=p/8V, the phase-mixer voltage V3 and the voltage V which is the vector sum of V1, V2 and V3. That is,

The significance of V3 in this equation may be illustrated by reference to a conventional feedback system operating without benefit of V3. The vector diagram of such a system is shown in Fig. 3, Where the amplitude and phase angle of a is the same as for Fig. 2. It is a well-known fact that if qb is 0 (or 360) and the ratio of V2 to V is greater'than unity, the conventional feedback system will become unstable; that is, it may self-oscillate. One purpose of my invention is to avoid this unstable operation or self oscillation in a negative feedback amplifier.

Again referring to Fig. 2 in which the magnitude and phase angle of a is the same as for Fig. 3, a stable condition of operation may be achieved by developing a voltage represented by vector Vs having a magnitude and phase angle for all frequencies such that, when added to voltage vector V2, a resulting or net feedback voltage vector V4 will be produced which will always be substantially from V1 and have an amplitude which is a xed fraction of thev amplitude of V1 for all frequencies. Such a system cannot oscillate, since for any value of i the net feedback voltage will always have substantially the amplitude and phase angie as that corresponding to operation at mid-frequency at which qs equals 180.

The manner in which the voltage represented .by vector Vs is developed will now be explained by referring to the vector diagram of Fig. 4. This figure represents the phase and amplitude relations between the voltages impressed upon the inputs of the phase mixer 28, the voltages developed therein, and the resultant or output voltage of the phase mixer. The nal (net) voltage, represented by vector V and impressed on the grid 4 of tube 5, is also impressed upon the grid I6 of amplifying tube I1 and upon the grid 25 of tube 26, the latter constituting one element of the phase mixer 28. The voltage impressed upon grid i6 appears, after amplification in the correlative amplifier, across resistor 22. This voltage will have a phase relation identical with that appearing across resistor il] in the output of the main amplier 9, it being remembered that the correlative amplifier is so designed that its gain and phase shift-frequency characteristic is substantially the same as that of the main amplifier. The contact 23 selects the desired magnitude of this voltage which is impressed on the grid 29 of tube 21, the latter constituting the second element of the phase mixer 28. This voltage is represented by vector Ve on Fig. 4.

There are now two voltages, V and Vc, operating on the phase mixer. The Voltage V is amplified and reversed in phase by the tube 26 and appears in the output circuit thereof between the point 40 and ground. This amplified voltage is represented by vector V5 in Fig. 4. Voltage Vc is amplified and reversed in phase by the tube 21 and also appears in the tube output circuit between the point 40 and ground. This amplified voltage is represented by vector Vc. The resultant of the voltage vectors V5 and Va is Va.

The phase angle of V3' is dependent on the mag-2 nitude of Vc which in turn is determined by the position of contact 23 on resistor 22. The magnitude of V3 is also dependent on the position of contact 23. This latter dependency is of particular importance for the condition where qs equals 180 as will be seen when the preferred circuit adjustments are described.

The voltage represented by V3 is amplified by the amplifier 4I and appears without change of phase across resistor 42. Contact 43 determines the magnitude of the voltage which is impressed back on the grid 4 of tube 5, and has already been defined as vector V3.

To facilitate the desired adjustment of contact 23 on resistor 22 it is preferable to make said adjustment when the applied voltage from source lf has the frequency for which =180, that is, mid-frequency. This frequency may be obtained by plotting a polar diagram of the magnitude and phase angle of a over a range of frequencies sufiiciently great to include the mid-frequency. Since, when employing this frequency, the voltage V3A should be zero, it is only necessary to place a suitable voltmeter across the output of the phase mixer and make the proper adjustment of contact 23 until this voltmeter reads zero. This follows from the fact that =180 and, in Fig. 4, Vc is in phase opposition to V.

The desired adjustment of contact 43 on resistor 42 may most readily be made by applying a voltage from source l having a frequency for which the phase angle qs is zero, that is, a critical frequency. It is preferable that the particular critical frequency selected be the next higher frequency above the mid-frequency. When employing this frequency, the voltage V3 should be of such magnitude as to reduce the output of the main amplifier to that value which corresponds to aVi (for the critical frequency) divided by the numerical value of l-a at mid-frequency. This follows from the fact that when using the correlative amplifier and phase mixer in connection with a feedback amplifier in accordance with my invention, the net gain of the feedback amplier should be a (measured at a particular frequency) divided by the numerical value of l-a measured at mid-frequency. A suitable voltmeter may be placed across resistor l and the position of contact 43 varied until the output voltage corresponds to that indicated above.

The value of this voltage can also be obtained by employing data obtained from the above described polar diagram provided the daigrain is extended to include a frequency for which =0. This will be clear by referring to Fig. 5 in which the curve represents the locus of a for a range of frequencies suiiicient to include a frequency for which q =180 and a frequency for which =0.

When adjusting the contact 23 on resistor 22 to obtain zero output from the phase mixer as above described, a given voltage V1 at the frequency for which =180 is applied between junction point 3 and ground. This voltage is not applied directly but is the result of a given voltage from source l and the known resistance values of resistors 2, I4, and 46. This given voltage V1 may be assumed to be unity. When the output of the phase mixer is zero the output voltage of amplifier 9, as measured across resistor I0 or between the contact l2 and ground, is noted.

When adjusting the position of contact 43 on resistor 42 to produce the required operating conditions, the frequency ofsource l is changed to the value for whichl ==0. From the diagram of- Fig. 5 the ratio o f the scalar value of a for =O to the scalar value'of a for =180 is obtained. With the voltage V1 at the frequency for which Il' 1*#13 that is, the output. voltageis lf V The vector relations for this circuit are shown in Fig. 3. It is apparent that V will not remain constant either in position or in magnitude for different frequencies, whereas in accordance with my invention the vector V remains constant both in position and magnitude for any value of a as shown by Fig. 2. The significance of this fact is that the gain and phase-frequency characteristic of the conventional negative feedback amplifier is altered by the feedback circuit, whereas, in accordance with my invention, the gain and phase-frequency characteristic is unchanged by the feedback circuit. In the latter instance, the gain is reduced by a constant proportion corresponding to the degree of feedback established at mid-frequency and the output voltage is aV.

So far nothing has been said regarding distortion voltages which may be developed in the main amplifier. It will now be shown that these distortion voltages will not produce self-oscillation and that without change in adjustments for the correcting Voltage V3 the phase of the final distortion voltages will always coincide with the phase of the distortion voltages as iirst applied to the grid 4 of tube 5.

'The distortion voltages first appear Within the circuits of the main amplifier and are impressed upon the output resistor l0. A portion of these voltages, depending on the position of contact l2, are fed back onto the grid 4. The voltages thus impressed on grid 4 will set up a system of voltage vectors in exactly the same manner as if the distortion voltages originated in the source l The operation of the correlative amplifier and the phase mixer circuits produce' the required correcting voltage and a vector diagram of the various voltages throughout the circuit resulting from the distortion components will be identicalwith that of Fig. 2, where V1 represents the contribution of the distortion voltage appearing in the amplifier output through resistor i4 to the net voltage V.

In order to show the extent to which distor` tion is suppressed in the circuit of my inven. tion reference is now made to the vector diagram of Fig. 6. In this diagram I have reproduced the vectors of Fig. 2 without change in nomenclature except to substitute for V the reference character D (for distortion).

In Fig. 6 the vector Di represents the distortion voltage at an assumed frequency as it first appears on, or is applied'to, the input of the tube 5 at the junction point 3, it being remembered that distortion voltages are rst generated within the ampiirler 9. In accordance with the explanation already given of the operation of the correlative amplifier and phase mixer, the result of this applied distortion voltage will be a feedback voltage D2 and a phase mixer voltage D3 resulting in a net voltage D. It is here assumed, for simplicity, that B has no inherent phase angle, so that for the distortion frequency assumed the phase angle o1 is the phase angle of u.

Vectors D1 and D2, although representing voltages between the junction point 3 and ground, are directly proportional to, and therefore representative of, the distortion voltage first appearing across output resistor Il] without feedback and to this voltage as modified by feedback, respectively. The resultant distortion voltage is therefore proportional to the vector sum of D1 and D2 or D. It will be seen, therefore, that in accordance with my invention the distortion has been changed in the ratio of vector D1 to vector D.

If the frequency of the distortion voltage developed Within the amplifier is such that the phase angle of fl=2 and the feedback voltage is represented by D2', the resultant distortion voltage is proportional to vector D, that is, the sum of vectors D1 and D2. In this case it will be evident that the final distortion with feedback is considerably greater than the distortion without feedback. In other words, in so far as distortion components for frequencies having considerable phase shifts are concerned, there may be no reduction in distortion due to feedback.

It may be well to point out that the vectors D2 and D2 have been drawn such that they have the same magnitude as vector D4. Usually attenuation accompanies increased phase shift with the result that in an actual amplifier the voltage vector D2 would be somewhat smaller than that of D3, and the voltage represented by vector D2 would be smaller than that of voltage vector Dz. If this were the case, the final distortion voltage vector D" would not be as great as that shown in Fig. 4. In any case, however, it is only the distortion components having small phase shifts which are considerably reduced by the circuit arrangements of my invention.

As shown in Fig. 6, if the feedback voltage maintains a constant Value regardless of frequency, the locus of the ends of the vectors representative of the final distortion is a circle shown by the dotted circle 41 having its center at the extremity of vector D1, and having a radius equal to D2.

One very objectionable distortion component encountered in practice is due to the use of alternating current for heating the cathodes of the vacuum tube amplifiers and to the ripples in the rectified voltages for supplying power to the tube anodes. These components usually occur at frequcncies which are subject to only a small phase shift. It will therefore be seen that the circuits of my invention are particularly valuable in reducing distortion components due to hum because an unusually great amount of feedback may be employed for reducing these distortion components without self-oscillation.

It may be desirable to more completely suppress a distortion component at some particuular frequency. This may be done by inserting a phase shifting network in the amplier circuits such that the phase angle of up will be substantially 180 at the desired frequency.

Another advantage of the circuit arrangements of my invention is that they may be applied to any existing amplifier whose operation it is desired to improve. All that is required is that a polar diagram showing the variation of ,u with frequency be made of the existing amplifier, and a correlative amplier be constructed which has a ,a-frequency characteristic represented by substantially the same diagram. The correlative amplifier 2 l, the phase mixer 28 and the amplifier 4| constitute an active network which may be connected to the circuits of the main amplifier at a single point, namely the junction point 3. There is no need of making any change to the amplifier circuits whose Iperformance is to beimproved other than, for example, the adding of isolation resistors such as resistors 2 and I4, and these may already be part of said amplifier.

Another manner in which the applied voltage, the feedback voltage, and the phase mixer voltage may be caused to act independently is shown in Fig. 7. In this ligure I have dispensed with the isolating resistors I4 and 4B and have submitted a multigrid tube 48 for the triode 5 of Fig. l. Resistor 2 may now have la relatively low value without affecting the operation of my invention. I have illustrated the tube 48 as having three control grids, 49, 5i] and 5I, one for each of the voltages which is to independently exert control on the anode current of the amplifying tube d8. A voltage representing the resultant of the combined effects of these voltages will be found to exist across the resistor 52 connected between the cathode 53 and ground. The desired potential to be applied to tubes l'l and 26 may be selected by contacts 54 and 55 respectively on resistor 52. It will be noticed that no phase change has been assumed between the-voltages applied to the grids of the tube 48 and those voltages obtained from the resistor 52. This will be substantially true for the range of frequencies of greater interest. Resistors 56 and 51 are grid leaks for establishing an appropriate D. C. potential on the grids 5| and 49 of the tube 48.

Other modifications of my invention will occur to those skilled in the art and it is to be understood that the illustrations given in this specication are given by way of example only and not as limiting the scope of the invention as set forth in the objects and the appended claims.

In certain of the claims I have used the expression vector-frequency characteristic. A vectorfrequency characteristic is understood to be the vector quotient of the output voltage divided by the applied input voltage over a frequency range. The vector-frequency characteristic takes account of the phase angle relationship as well as the amplitude relationship between the output Voltage and applied input voltage.

What is claimed is:

1. In combination, a feedback amplifier having an input circuit and an output circuit and a feedback connection therebetween, a correlative amplilier having an input circuit and an output circuit, and a. phase mixer having two input circuits and an output circuit, means connecting the input circuits of said feedback land correlative amplifiers and one of the input circuits of said phase mixer together at a common point, means connecting the output circuit of said correlative amplifier to the other input circuit of said phase mixer, means connecting the output of said phase mixer to said first connecting means, a source of potential applied to said common point, and a load circuit connected to the output circuit of said feedback amplifier.

2. The combination of claim 1 in combination 11 -With means for adjusting the output voltage of said phase mixer to zero for a frequency of said source of potentia1 for which phase shift through said feedback amplifier is 180.

3. In combination, a feedback amplifier having an input and an output circuit and a feedback connection therebetween, a correlative amplifier having an input and an output circuit, a phase mixer having input and output circuits, means connecting the input circuits of the correlative .amplifier and phase mixer to the input circuit of the feedback amplifier, means connecting the output circuit of the correlative amplifier to the input circuit of the phase mixer, means connecting the output circuit of the phase mixer to the input circuit of the feedback amplifier, a source of applied voltage having a range of frequencies connected to the input circuit of th'e feedback amplifier and voltage adjusting means for adjusting the magnitude and phase of the voltage developed by the phase mixer to a Value such that the final voltage effective in the input circuit of the feedback amplifier will have the same phase as the applied voltage from said source over said range of frequencies.

4. The combination of claim 3 in which the in' put circuit of the feedback amplifier comprises a multigrid tube, the applied voltage being confnected to a grid cf said tube, said feedback connection being connected toa grid of said tube; and the output of the phase mixer being connected to a grid of said tube.

5'.` A negative feedback amplifier comprising an input circuit and an output circuit and. al feedback connection therebetween for provid-ing a 35 2,244,249

feedback voltage and an additional circuit connected across said input circuit for producing an additional voltage which when combined with said feedback voltage will produce negative feedback over a wide band of frequencies, said additional circuit comprising a phase mixer circuit having two input circuits and an output circuit and a correlative amplifier having an input and output circuit, said input circuit of said correlative amplifier and one of said input circuits of said phase mixer being connected to the input of said amplifier, the output circuit of said correlative amplifier being connected to the other input circuit of said phase mixer, and said output circuit of said phase mixer being connected to said input circuit of said amplifier.

6; A negative feedback amplier according to claim 5, further comprising adjustable means connecting said correlative amplifier to one of the input circuits of said phase mixer for adjusting the magnitude and phase of the voltage applied thereto such that the output Voltage of said phase mixer is zero for frequencies for which the phase shift of said feedback voltage is HUGO ROMAN'DER.

REFERENCES' ciED The following references are of record in`` th file of this patent:

UNITED STATES PATENTS Number Name Date n, 1,994,486 Robertson Mar. 19, 1935 2,227,048 West Dec. 31, 1940 Guanella June 3, 1941