US1894322A

US1894322A - Means for eliminating distortion in repeaters

Info

Publication number: US1894322A
Application number: US393241A
Authority: US
Inventors: Nyquist Harry
Original assignee: American Telephone and Telegraph Co Inc
Current assignee: AT&T Corp
Priority date: 1929-09-17
Filing date: 1929-09-17
Publication date: 1933-01-17
Anticipated expiration: 1950-01-17

Description

Jan. 17, 1933-5 y u g-r 1,894,322

MEANS FOR ELIMINATING DISTORTION IN REPEATERS Filed Sept. 17. 1929 Phase Retard v 2 I" 8 5 7' INVENTQR q BY E/lQguw ATTORNEY Patented Jan. 17, 1933 V UNITED STATES PATENT-OFFICE HAD-BY NYQ'UIBT. O1 MILLBURN, NEW JERSEY, ASSIGNOB TO AMERICAN TELEPHONE AND TELEGRAPH COMPANY, A CORPORATION OF NEW YORK MEANS FOR ELIHINATING DISTORTION IN REPEATEBS Application filed September 17, 1989. Serial No. 393,241.

This invention relates to wave translation systems, as, for'example, systems for amplif ing electrical variations with the aid of e ectrical space charge devices, and, more particularly, it relates to improvements in such systems and in the circuit arrangements of the space charge devices therein.

Re eaters have been proposed for neutralizlng distortion by picking oil some of the output wave from the output of an amplifier which distorts, neutralizing the undistorted portion of the wave thus picked oil? by means of an undistorted wave from the input of the amplifier, and then impressing the resultant wave, which comprises distortion components only, upon the input of the amplifier in such phase and amplitude as to neutralize the distortion due to the amplifier. Such a system, if properly adjusted will not only be substantially distortionless, but will also compensate for variations in the gain of the amplifier due to variations of its circuit constants, and will thus tend to maintain the gain uniform.

In order that the repeater be effective to produce these results, the efi'ect of reactance elements in the repeateritself should beneutralized or compensated, particularlycapacities between the grids and plates of the vacuum tubes and the capacities to ground of the interstage wiring and coupling elements which are in effect bridged ca acities which tend to cause phase shifts wh1ch vary with frequency.

In accordance with the present lnventlon, it is proposed to neutralize or compensate such capacity reactances by bridging across each amplifying element a circuit including an inductance in series with resistance so proportioned with respect to the inherent bridged capacity of the circuit as to neutralize the capacity so far as effects involving the first order offrequency are concerned. Where necessary to compensate for eifects of the second order or higher powers, an auxiliary network may be provided at some point in the system having such characteristics that its impedance will be real so far as concerns terms lower than the second order; while for terms of the second order (or even higher powers) its impedance will be complementary to the impedance of the circuit as previously compensated. v

The invention will now be more fully understood from the following description when read in connection with the drawing, in which Flgure 1 is a schematic diagram of a type of repeater to which the invention may be applled; Figs. 2 and 3 are diagrams illustrating arrangements for compensating the effect of reactance in accordance with the invention; Fig. 4 is a diagram used. in connection with the derivation of certain formulae; Figs. 5,- 6 and 7 show amplifier circuits compensated in accordance with the invention; 65 Fig. 8 is a detailed circuit diagram of a multlstage distortion neutralizing repeater to which the compensation in accordance with the invention is applied, and Fig. 9 is a series of curves illustrating the principles of the inventlon.

While the methods of com ensating the eifects of reactances in ampli ers as herein disclosed are of general application, they are particularly useful in connection with repeaters of a type arranged for neutralizing distortion. In order to understand the necessity for the invention in connection with such a repeater and how it is to be applied thereto, it is desirable to describe briefly the principles of the repeater to which the invention is applied. i

A distortion neutralizing repeater is illustrated schematically in Fig. 1 and comprises amplifier elements A A A and A connected as shown, together with an attenuation equalizer E to compensate for the variation in the attenuation of the currents with frequency as they are transmitted over the line. In this typelof repeater, the amplifying-unit A is the principal amplifying unit for bringing the currents up to a desired large volume,.the amplifying units A A and A, being worked at such low levels that the'nonlinear distortion due to their action may be neglected. Due to the high levels atwhich the amplifier A must be worked, said amplifier will have non-linear distortion which it is the function of the amplifiers A and A 1 0 with their associated circuits M, N and S, to compensate.

Let us assume that a pure sinusoidal wave is applied to the input of the repeater and that the amplifier A is not functioning. A pure sine wave will then appear at the point between amplifier units A and A On the output side of amplifier A a similar sine wave appears, but it has superposed thereon certain distortion currents. By means of the resistance coupling X X and X the c1rcuit M picks off a portion of the out ut wave with its distortion, and, by means 0 the circuit N, the wave picked off is combined with a portion of the wave appearing between the amplifiers A and A If care is taken that the phase and relative amplitudesof these two waves are properly related, the smusoldal wave coming over the circuit N from the point between amplifiers A and A will just neutralize the sinusoidal part of the wave coming from the output of the amplifier A As a result, the current which is applied to the input of the amplifier A is made up of the distortion only.

Now, if the amplifier A is made to function, this distortion will be introduced into the main circuit atthe point between the equalizer E and the amplifier unit A Accordingly, it will be amplified by amplifiers A and A, and will appear in the output. If care is taken to properly relate the am litude and phase of this distortion current which was applied to the circuit over the path S) to the phase and amplitude of the distortion components of the sine wave after transmission through the amplifier A this distortion may be made to neutralize the distortion generated in the amplifier A so that the ultimate output of the am lifier A will be an undistorted sine wave. 11 the other hand, if the gain in the amplifier A is made somewhat too great (or too small) the repeater may be made to overcorrect (or undercorrect) for the distortion, so that, for instance, distortion generated in the amplifier A or A may be compensated, assuming, however, that amplifiers A and A have the same proportion of sgcond and third harmonics as the amplifier A repeater such as above described has another desirable property in that it tends to compensate for variations ingain. For example, suppose the gain in amplifier A changes for some reason, such as a variation in battery Voltage. As a result, the fundamental wave is not completely neutralized before entering the amplifier A In other words, a portion of the fundamental wave is amplified by the amplifier A, as well as the distortion. The net result of this fundamental wave component, when applied to the amplifier-A is to keep the output into the line constant. For example, if the fundamental component coming from the circuit M is greater than that coming from the; circuit N, the resultant wave passing through the amplifier A will be in such phaseas to oppose the excess transmitted' from the output of the amplifier A to the circuit M. On the other hand, if the fundamental component coming from the circuit M is smaller than that coming from the circuit N, the phase of .the'resultant wave passing through the amplifier A, will besuch as to add to' the fundamental component appearing in the output circuit of the amplifier A In other words, the circuit not only tends to keep the distortion zero, but tends to keep variations in the gain of the amplifier A from having. an effect on the over-all gain.

In order to have proper functioning of this type of apparatus, it is necessary that the phase relations between the various waves be approximately right. This leads to the following requirements:

1. The amplifiers A A and A must be resistance coupled, or very nearly so.

2. The total number of stages in amplifier A must be an odd number.

3. The total number of stages in amplifier A, plus amplifier A must be an even number.

In considering the various component amplifiers, it should be noted that the ampliequalizer in front of the repeater, because the resistance noise originating in the equalizer is superposed on that coming from the line. The effect of the amplifier A therefore, is to increase the power of the incoming current so that the effect of the resistance noise introduced by the equalizer will be made relatively unimportant. This amplifier does not have the benefit of the distortion correcting properties of the circuit for the reason that there will be unavoidable phase shifts in the equalizer. tion introduced in the amplifier A must, therefore, be kept small.

The impedance looking from the amplifier A towards the equalizer E should be a'pure resistance in order to insure resistance coupling between the amplifiers A and A This result can be readily brought about by properly choosing the design of the equalizer. The equalizei' itself may be of any type well known in the art and is not inherently a part Any distoramplifier A should be odd, and the total number in A, and A even. Any number necessary to obtain the desired gainwillbe.

amplifier A, and added to u'se'd. Of course, stages ina be taken from 2 so long as the number in A, remains odd, and the sum of the stages in A, and A, is even. There is a distinct disadvantage in this, however, 1n that the stages in the amplifier A: will not be within the compensated portion of the circuit, and, therefore, an distortion generated in these sta es woul be imperfectly compensated. urthermore, variations in the gain of the amplifier A would not be properly compensated. In fact, the stages of the amlifier A may all be embodied in the amplier A provided the number of stages is even. On the whole, the preferred design would appear to be, to have a limited even number of stages in amplifier A eliminatedentlrely, so that the paths S and N will be bridged together across the coupling between the equalizer E and the amplifier In the detailed design given hereinafter, this modified form of circuit is utilized.

Inorder for the type of repeater under discussion to work at its best, it is necessary that the phase shift due to reactances be kept at a very low value. This is desirable not only for some one frequency but for the whole range of frequencies which it is wished to amplify. The circuits involved 1 n vacuum tube coupling are of much higher impedance than the natural impedance of the wiring used. Since, for capacity, the phase shift is proportional to and, for inductance, the phase shift is proportional to the stray capacity is much more likely to cause undue phaseshift than stray inductance.

There are three principal sources of capacity which interfere with the ideal operation of the circuits. First, there is the matter of unavoidable capacity to ground of the resistances used for the resistance coupling between stages. Secondly, there is the ca pacity to ground of the leads connecting the plate of one tube to the grid of the next. In this class, should be included the capacity to ground of the series condenser which is used in this lead. Third, there are the inter-electrode capacities in the tube itself, of which the capacity between the grid and the plate is particularly objectionable because its action is multiplied by the amplification factor of the tube.

Although it is possible to compensate for these capacities it is desirable to keep them at a low value; first, because it is inherently impossible to obtain perfect compensation and the smaller the capacities are to begin with the smaller will be the 'uncom ensated part, other things being equal. econdly, there are unavoidable variations from one tube to another and it is, of course, more difficult to compensate for capacities which are 1 of a variable nature than those which stay fixed. In addition there are variations in tube resistance which would alter the phase shift even if the capacity did not change. It is desirable then to keep variations small and it is thought that one way of accomplishing this is to make the capacities small to be in with.

he series condenser which is used between the plate of-one tube and the grid of the next should preferably be enclosed in a metal container to avoid crosstalk from stray fields. ,Unless special care is taken, the capacity between the condenser proper and the container will be great enough to have serious efiects. This capacity can be kept within a few micro-farads by keeping the dimensions of the condenser fairly small and keeping the separation from the condenser proper to the container of the same order of magnitude as the dimension of the condenser.

The leads between the plate of one tube and the grid of the next should have small capacity. This may be ensured by making them short, giving them a small diameter and keeping them far away from grounded parts.

The resistances which supply grid and plate potentials to the tubes should have as little distributed capacity as possible. This can be accomplished by making them of as small dimensions as possible particularly in crosssection and keeping them as far from ground as possible and keeping the high potential leads short.

The tube capacities and tube resistance should be made as small as possible. The tube resistance can be made low byusing such potential on the grid and plate as to work the tube on the steep part of its characteristics. Another expedient is to use tubes of small dimensions. If of two given tubes one is a small scale model of the other the amplification factor and resistanceare said to be the same for the two tubes whereas the capacities are directly proportional to the 7 linear dimension.

The capacities which have been discussed are in the nature of bridged capacities, and their effect is to retard the phase. This is also true of the capacity between the plate and the grid, although it. is not strictly a bridged capacity. Now, series capacity has the opposite effect. That is to say, it tends to advance the phase. Unfortunately, however, it is not suitable as a compensating means, because it has a totally different frequency characteristic from what is desired. The retarding effect of bridged capacity increases as the frequency increases, as shown by the curve a of Fig. 9, whereas the advancing effect due to a series condenser decreases as the frequency increases, as shown by the curve I) of Fig. 9. If the series condenser were so chosen as to compensate perfectly at some particular frequency, the compensation would be very imperfect at frequencies above and below the chosen frequency.

Bridged inductance has the same effect as series capacity, but it also has the same sort of frequency characteristic as shown by the curve I), and, therefore, is no more suitable as a compensating means than series capacity. Furthermore, a bridged inductance would inevitably introduce bridged capacity due both to the capacity between the turns and the capacity to the core and container.

In accordance with the present invention,

however, it is proposed to overcome these difficulties by neutralizing the effect ofthe bridged capacities by means of a bridged circuit in which a small inductance is in series with a resistance. Notwithstanding that a pure bridged inductance produces a variation of phase advance with frequency, as shown by the curve I) of Fig. 9, a combination of inductance and resistance, when properly designed as hereinafter described, will introduce a phase change with frequency which is complementary to that due to the bridged capacity as shown by the curve d of Flg. 9. This compensation may be made accurate so far as first-order effects are concerned, although, as will be pointed out later, further compensation may be made by special networks for second power or other higher order effects, if desired. In using a bridged circuit comprislng an inductance in series with the resistance, there will, of course, be a certain amount of difliculty due to stray capacity in the coils, but it should be noted that the E. M. F. across the coils is relatively small, since most of the drop in potential will take place in the resistance in series with the coil, which will be used to supply the grid and plate batteries. For this reason, stray capacities in these coils are not so objectionable.

In order to understand how the. combination of a resistance in series with an inductance bridged across the circuit will have the effect of compensating for a bridged capacity, let us suppose we have an effective bridged capacity C, as shown in Fig. 2, and it is desired to neutralize the phase change due to such capacity. The capacity C may be assumed to be any of the various capacities heretofore discussed in connection with a vacuum tube, for example, the grid-plate capacity. The proposition is to neutralize the efiect of such a capacity by means of a small inductance L in series with a resistance R, the whole being bridged across the circuit, as shown in Fig. 2.

That such an arrangement will have the desired effect will be apparent from the following considerations:

A network, such as shown in Fig. 2, will lntroduce no out-of-phase component if its impedance is real, as the absence of any imaginary factor indicates that the reactance effects are neutralized. Bearing in mind i and that of is z'wL, where a) is 211' times the frequency, the 1mpedance Z of the combination may be written 1 Z (R wL) 1 R 'LwL which by simple algebraic transformation becomes that the impedance of C is Now, if in equation (2) we neglect second order effects (those involving powers higher than the first power of the frequency) the 1mpedance may be written it +'iwL 1+T0R (3) But from inspection of the foregoing, this impedance Z will be made real if we make L of such value that the following proportion holds:

Rzz'oLz zlzz'waR (4) for, if this proportion holds, the imaginary operator i cancels out. Incidentally, the frequency factor 1 also disappears. Solving proportlon (4), we have R as the condition for making the impedance real so far as effects of the first order are concerned.

If, now, we substitute the value of L given by equatlon (5) in equation (2), we may wrlte Analysis of equation (8) shows that the 1mpedance is real and independent of frequency for all terms lower than those involving the second power of frequency or higher powers, being in fact equal to R. It should also be noted that the second power term w G R is also real though not independent of frequency. Computations for practical cases indicate that thirdpower and higher terms are negligible andneed not be compensated for frequencies up to 100,000 cycles.

' The effect of this second power term should not be neglected, however. While this term is real and hence involves no phase shift, its effect in an amplifier system is to increase the gain slightly with frcquenc This effect can be compensated, as will he described later, by introducing an auxiliary network of such character that it introduces no first order distortion but for the second power term produces again characteristic which decreases with frequency.

Before proceeding, with the consideration of the design of such auxiliary network, let us consider how the principles so far developed may be applied to the design of an amplifier. One form of application is illustrated in Fig. 5. Here, the grid biasing potential is supplied from a grid battery through an inductance L self-inductance of a winding of a transformer L and the usual battery supply resistance R The plate battery will be supplied through an inductance L and the self-inductance of the other winding of the transformer L in series with the plate supply resistance R A condenser of large capaclty is introduced inv the lead between the pla'te of one tube and the grid of the next tube between the battery supply resistances, but this condenser is of very lar e capacity, for the purpose of insulating t e grid and plate circuits with respect to the direct current supply, and does not introduce any material efi'ect from. an alternating current standpoint. The same holds true of the two condensers associated with the filament, for the purpose of connecting the grid and plate to a neutral point in the filament circuit. The effective bridged capacities which are to be neutralized are principally the capacity from grid to plate and the various capacities to ground elsewhere referred to.

The coil L represents a transformer of per-,

feet coupling whose mutual inductance and self-inductance per winding equal L This coil compensates for the grid plate capacity C The coil L compensates for the capacity G which is made up of the capacity from 'the'grid to ground and filament plus incidental capacity in the grid wire. L compensates for G which is the capacity of the plate and its Wire. puting the; inductances are:

' Where R is the resistance connected to the grid, and R isthe resistance connected to the plate.

Equations (9) and (10) follow obviously from equation (5). In order to understand The formulae for com-.

the derivation of equation (11), however consider Fig. 4 which is Fig. 5 in simplifie form, the inductances L and L bein omitted as their effect in compensating or the capacity C is negligible. Assume that in Fig. 4 a voltage is a plied to the plate. Then, current I flows rom plate through 0 R, and L The impedance of L isnegligible in comparison with R and G Hence, impedance Z of the circuit is and the current flowing through such impedance is The drop across L is negligible. Hence, the drop e across R is This current through primary of L induces drop e across secondary of L which will be positive or negative according to the way transformer is poled. It may be written E'iwL EL]; 6 tawny i R.

If (3 is neutralized, there is no change potential of the grid dueto E. Hence e+e' must equal 0 and L must be poled so that e which was to be proven.

A modified arrangement is shown inFig. 6, and the theoretical considerations already given apply equally to this figure. Fig. 6 differs from Fig. 5 in that the terminal of the inductance L is connected .to the grid lead through a condenser, and the biasing poable to keep the Rs great, as any bridge across the circuit should be of high impedance in order to prevent shunting energy from the succeeding tube. The arrangement shown in Fig. 7 illustrates how two resistances may be combined for alternating current purposes. Here, the resistances R and R are connected in parallel in a bridged circuit including windings of transformers L associated with two succeeding vacuum tubes and a common inductance L The interstage capacities shown between the resistances R and R may be neglected in considering alternating current. It will be seen, therefore, that the two resistances in parallel may be used in computing the Us. The formulae now are:

where C is the total capacity to ground of the interstage wiring including the grid plate capacity, and R is the combined resistance of the plate and grid supply resistances R and R in parallel. The method of compensating shown in Fig. 7 is applied to a large number of stages in a complete distortion neutralizing repeater illustrated in Fig. 8, as will be described later.

The compensation so far discussed is of the first order in frequency. The second order. (that is terms involving the second power of frequency), is not compensated. The efiect of the second order term (m C R of equation (8) is to increase the gain slightly as the frequency increases within the frequency range of interest. Computations indicatethat with existing vacuum tubes it will be desirable to compensate for this second order eflect as well as the first in order to operate up to about 100,000 cycles. This apparently need not be done at 'each stage, however, if the precautions above enumerated are observed. A parently, the third order term iw CR-* of equation (8) need not be compensated under these conditions.

In accordance with the present invention, the second order effect may becompensated by introducing at some point in the circuit a net-work of the type shown in Fig. 3, which comprises an inductance e in series with a shunt combinatlon comprlslng a resistance aand a capacity 0. Such a network, in order to perform its function, should produce no first order distortion, and the second order term of its impedance should be complementary to the term involving the second power of frequency in equation (8) The impedance of a network such as shown in Fig. 3 may be written rzinezzlzaucr In order to determine the effect of the assumed value of e upon the higher power terms of the impedance, let us substitute the whence,

value of e given in equation (26) in equation (23), so that 1 +iwcr 2 2 2 Expanding equation (27), we have '=1"(1w 0 r +z'o c r (28) Comparing equation (28), with equation (8) it will be seen that the impedance of the network of Fig. 3, so far as first order effects are concerned, is, like that of Fig. 2, a pure resistance r, and its second order term m 0 7' is opposite in sign but otherwise similar to the corresponding second order term of equation (8) Obviously, then, if the elements 6, c and 1' of the network of Fig. 3 be given the values L, C and B, so that L=CR the impedance of the network of the Fig. 3 to the second order may be written The second order term of equation (29) 'w CLR will annul the corresponding term 10 0 13 +m CLR of equation (8) Further comparison of equations (8), and (28) shows that the terms of the third order of frequency are of similar form but opposite in sign, so that if a network such as that of Fig. 3 be provided for each stage of the amplifier, the effect of the reactance could be compensated for all terms up to and including the term involving the third power of frequency. Figure 8 shows a distortion neutralizin repeater of the type shown in Fig. 1, to w ich the reactance neutralizing networks of Figs.

2 and 3 have been applied. In this figure, the

amplified A is represented by a single stage vacuum tube LT The amplifier A has five stages, and the amplifier A, has two stages. The phase compensating member, as shown in Fig. 2, is applied to each ofthe stages of the amplifiers A and A. in the form shown in Fig. 7, and, therefore, for each stage of each of these, the phase retarding effects of the grid-plate capacities, as well as the capacities of the grid wiring to ground and the plate wiring to ground, are compensated to the first order by the use of the coils L and L in combination with the grid and plate su ply resistances connected in parallel.

econd order efi'ectsin the amplifier A are compensated by introducing the network N between the third and fourth stages, as shown, and the same result is accomplished for the amplifier A, by introducinga similar network N in the grid circuit of the first stage.

The filaments of all of the tubes of the three amplifiers of Fig. -8 are connected in series and supplied from a twenty-four volt battery, as shown. In order that the grid biasing potentials may be supplied from the filament circuit, it is desirable to arrange the order of the filaments in the filament supply circuit, so that, beginning from ground, the filaments of the tubes come in the following order: First, the filament of the amplifier A followed by the filaments of the five tubes of the amplifier A in reverse order, beginning with the last stage tube LT and ending with the first stage tube NT.,, and these in turn being followed by the filaments of the tubes NT and NT, of the amplifier A The reason for this is that in the particular design illustrated, the tube of the amplifier A is an L tube, and, likewise, the last stage of the amplifier tube A has an L tube LT while all of the other tubes of the amplifiers A and A, are so-called N tubes.

The drop through the filaments in each of the L tubes is about four volts and that through each of the N tubes is about one volt. Each of the N tubes should have its. grid biased about six volts negative with respect to the negative side of its filament, while the grids of the L tubes should be biased about nine volts negative with respect to their filaments. In order to give the proper grid taps for each of the tubes, therefore, the filaments of the tubes are arranged in the order above stated, and additional resistance dro s of about one volt each, numbered 9 to 14, 1116111- sive, are provided. If the connection of each grid to a tap in the filament supply circuit is traced, it will be found that each L tube will have its grid connected about nine volts more negative than its filament, and, accordingl each N tube will have its grid connected a out six volts more negative than its filament. I

In connection with the amplifier of Fig. 8, the method of obtaining the feed-back volta e into the amplifier A. from the last stage of the amplifier A, should be noted. The output impedance of the last stage tube LT: and the impedances X X and X form the four sides of a Wheatstone bridge. The feed-back voltage impressed across the grid and filament of the first sta e tube NT; of the amplifier A; is picked 0 across one diagonal of the bridge, namel the junction point of the resistances X an X and the junction point of the resistance X with the filament plate impedance of the tube LT The output transformer T is connected across the opposite diagonal of the bridge. It follows, therefore, that the transformer and the line into which it works are in conjugate relation to the amplifier A so that the transformer does not introduce any phase shift in the current fed back to the amplifier A... Moreover, the performance of the circuit will not be affected by changes in the line impedance.

The compensating networks N and N of the amplifiers A and A; should preferably be made adjustable in order that they may be readjusted to take care of the changed characteristics of the amplifier when tubes are replaced. In order to readjust the network N the key K is thrown to the meter M thereby opening the circuit of the amplifier A whichis now disabled. When theamplifier A is operating normally, there, should be no gain in the local round .trip circuit through this amplifier, beginning with the connectionN, through the amplifier and over the output connection S back again to the inut connection N. The reason for this may 2 made clear if we refer again to Fig. 1 and consider the local round trip circuit including amplifier A path N, amplifier A; and

path S. It will be remembered that amplifiers A and A in this diagram are combined in the one amplifier A in Fig. 8, but the rinciple of operation is the same. Rememliering that the function of the amplifiers A and A is to bring the distortion component from M back to the input of the amplifier A with such phase and amplitudeas to neutralize the distortion in the output of A,., it is evident that if there were any gain in the round trip circuit through the two amplifiers A and A the distortion component from M would be over-compensated; and hence the normal adjustment of the amplifier A, of Fig. 8 is one in which there will be no gain in the round trip circuit through this amplifier, but, on the contrary, the amplifier will be substantially at its singing point. Therefore, when the amplifier A is opened by throwing the ke K the amplifier A if properly adj uste should be ust on the verge of slnging. If any singing occurs, this fact will be indicated on the meter M The network N should be readjusted so that the amplifier A is just at the singing point for each frequency over the range involved. To make this adjustment, the key K may be thrown to a variable oscillator 0. As the oscillator O is adjusted to each frequency, the fact that the amplifier A at that frequency is at a near-singing condition will be indicated on the meter M and the network N can be adjusted accordingly.

In order to adjust the amplifier A by means of the network N key K should be restored and key K thrown to the meter M The latter key opens the output of the amplifier A The gainfrom the input of the amplifier A through said amplifier and over the feed-back connection to the input of the amplifier A should be zero, because at this point, referring again to the-operation previously outlined for Fig. 1, the pure sine wave from the output of the amplifier A is met by a pure sine wave from the input of the amplifier A of opposite sign and equal mag nitude. If, therefore, the key K is thrown to the meter and the amplifier A is properly adjusted, nothing but distortion should appear upon the meter M The fact that the amplifier A is properly adjusted will be indicated by a minimum indication of the meter M By throwing the key K and connecting various frequencies from the oscillator to the amplifier A the network N may be adjusted in order to introduce minimum current on the meter M in which case the amplifier A will be adjusted for proper compensation.

It will be obvious that the general principles herein disclosed may be embodied in many other organizations widely different from those illustrated, without departing from the spirit of the invention, as defined in the following claims.

What is claimed is:

1. In an amplifier circuit in which an effective bridged capacity C exists, means to compensate first order of frequency effects of said capacity comprising a series combination of inductance L and resistance R shunted across said circuit, the resistance B being of appreciable magnitude as compared with the inductance L.

2. In an amplifier circuit in which an efl'ec tive bridged capacity C exists, means to compensate first order of frequency effects of said capacity comprising a series combination of inductance L and resistance R shunted across said circuit, said inductance and resistance being related to said effective capacity substantially in accordance with the formula L==UR 3. In an amplifier circuit in which an effecin uctance L and resistance R shunted across said circuit, said inductance and resistance being so related to said effective capacity that the lmpedance of the combination will be real and independent of frequency for terms involving powers as high as the first power of frequency.

5. In an amplifier circuit in which an effective bridged capacity C exists, means to coni- I pensate first order of frequency effects of said capacity comprising a series combination of inductance L and resistance R shunted across said circuit, said inductance and resistance being so related to said eflective capacity that the impedance of the combination will be real for terms as high as the second power of frequency and will be independent of frequency for terms below the second power of frequency. I

6. In an amplifier circuit in which an effective bridged capacity (1 exists, means to compensate first order of frequency effects of said capacity comprising a series combination of inductance L and resistance R shunted across said circuit, and means to compensate second order of frequency effects of said capacity comprising a bridged network including an inductance e in series with a shunt combina-' tion of a resistance r and a capacity 0.

7. In an amplifier circuit in which an effective bridged capacity C exists, means to compensate first order of frequency efl'ects p of said capacity comprising a series combination of inductance L and resistance R shunted across said circuit, said inductance and resistance being related to said efi'ective capacity substantially in accordance with the formula L=UR and means to compensate sec ond order of frequency effects of said capacity comprising a bridged network including an inductance e in series with a' shunt combination of a resistance r and a capacity 0, the elements of said network being related substantially in accordance with the formula 6 07 8. In an amplifier circuit in which an effective bridged capacity C exists, means to compensate first order of frequency effects of said capacity comprising a series combination of inductance L and resistance'R" shunted across said circuit, the impedance inductance L and resistance R to the second order of frequency being R+m ORL, and means to compensate second order of frequency efi'ects'o'f said capacity comprising a bridged network includin an inductance Z in series with a'shunt com ination of a resistance r and a capacity 0, the impedance of the network to the second order of frequency being r-acrl.

9. In an amplifier circuit in which an effective bridged capacity 0 exists, means to compensate first order of frequency effects of said capacity comprising a series combination of inductance L and resistance R shunted across said circuit, said inductance and resistance being so related to said'efifective capacity that the impedance of the combina tion will be real for terms as high as the second power of frequency and will be independent of frequency for terms below the second power of frequency, and means to compensate second order of frequency effects of said capacity comprising a bridged network including an inductance e in series with a shunt combination of a resistance r and a capacity 0, the elements 6, c and 1' of the network being so related that the impedance of the network to the first order of frequency will be real and independent of frequency and the second power term of its impedance will be complementary to that of the combination comprising L, C and R.

10. In a multi-stage amplifier circuit including a plurality of vacuum tubes each having a grid to plate capacity C and ca pacity C from the interstate wiring and apparatus to ground, means to compensate first order of frequency efiects of said capacities comprising a bridged circuit including grid and plate battery supplv resistances having a combined resistance R in series with the mutual inductances L of transformers for coupling the plate of a given tube to the grid of the same tube and a self-inductance L 11. In a multistage amplifier circuit including a plurality of vacuum tubes each having a grid to plate capacity C and capacity C from the interstage wiring and apparatus to ground, means to compensate first order of frequency effects of said capacities com: prising a bridged circuit including a grid and plate battery supply resistances having a combined resistance R in series with the mutual inductances L of transformers for coupling the plate of a given tube to the grid of the same tube, and a self-inductance L said mutual inductance having substantially the value L12: 0 59 and said self-inductance having substantially the value L 0 13 12. In a multi-stage amplifier circuit in cluding a plurality of vacuum tubes each having a grid to plate capacitV C12 and capacity C from the interstage Wiring and apparatus to ground, means to compensate first order of frequency effects of said capacities com rising plate of a given tube to the grid of the same tube, and a self-inductance L said inductances and resistances being so related to the corresponding capacities that the impedance of the combination comprising grid-plate capacity, battery supply resistance and mutual inductance L and also the impedance of the combination comprising interstage capacit to ground, battery supply resistance and sel inductance L will each be real and independent of frequency for terms of the first order of frequency.

13. In a multi-stage amplifier circuit including a plurality of vacuum tubes each having a grid to plate capacity C5 and capacity C from the interstage wiring and apparatus to ground, means to compensate first order of frequency effects of said capacities comprising a bridged circuit including grid and plate battery supply resistances having a combined resistance R in series with the mutual inductunces L of transformers for coupling the plate of a given tube to the grid of the same tube, and a self-inductance L said inductances and resistances being so related to the corresponding capacities that the impedance of the combination comprising grid-plate ca .pacity, battery supply resistance and mutual inductance L and also the impedance of the combination comprising interstage capacity to ground, battery supply resistance and self-inductance L will each be real and independent of frequency for terms of the first order of frequency and will each be real for second power terms.

14. In a distortion neutralizing repeater, a principal multi-stage amplifier, a feed-back connection from the output of said principal amplifier to its input through an auxiliary multistage amplifier, and a connection from the input of said principal amplifier to said auxiliary amplifier, each stage of each amplifier, having inherent capacities effectively bridged across the circuit, means to compensate first order of frequency effects of said capacities comprising a circuit bridged across each stage and including series inductance and resistance related substantially according to the formula L 013 and means to compensate second order of frequency efi'ects comprising a bridged network for each ampliier, said bridged network comprising an inluctance e in series with a shunt combina- Lion of capacity 0 and resistance 1' propor- LlOIlQCl substantially in accordance with the formula e=or 15. A path for transferring waves of a substantial frequency range, comprising shunt resistance and effective shunt capacity, and impedance means forming with said resistance and capacity a shunt impedance that is substantially pure resistance mpedance over said frequency range.

16. In combination, a transfer path for waves representing a signal, a reactance included in said path, an impedance network shunted across said path at a point between the ends of said path, reactances included in said network, the impedance of said network and ath between their junction points being a substantially pure resistance over the frequency range of said waves.

17; A one-way wave propagating path having an input section and an output section a network with its input impedance shunted across said input section and shunted across said output section and presenting to each of said sections approximately a pure resistance impedance over a substantial frequency range.

18. In combination, a source of waves and two wave transfer paths connected to said source in parallel relation to each other, one of said paths comprising a filter capable of selectively transmitting with substantially uniform low attenuation waves of a band of frequencies of substantial width which has one end connected to said source, said filter having a shunt condenser included in its initial impedance element at said one end.

19. Two vacuum tubes and a circuit for coupling them in tandem relation, said circuit forming with inter-electrode capacities of said tubes a network that has two input terminals, one connected to the cathodes of said tubes and the other connected to another electrode of each tube, the impedance presented by said network to said terminals being approximately pure resistance impedance over a substantial frequency range.

20. A path for transmitting waves, comprising vacuum tubes in tandem relation, im-

pedance means forming with interelectrode capacities of said tubes a network, a circuit for supplying space current to one of said tubes, a resistor included in said circuit and forming a terminating impedance for said network, said network having an input impedance which is substantially pure resistance impedance for said waves and which faces two of the electrodes of each of said tubes.

21. A wave transmission system comprising two vacuum tubes and coupling means for coupling said tubes in tandem relation for transmission of waves from one to the other,

aseaaaa network shunted across said ath at a point between the ends of said pat reactance included in said network, the impedances of said network and path between their junction points being a substantially pure and constant resistance overthe frequency range of said waves.

23. A system comprising a plural odd number of electric space discharge devices, interstage coupling circuits for said devices, means for balancing waves from the output side of the last device against waves from the input side of the first device, means for feeding waves resultin from said balancing action through said evices, and phase correcting megns in each of said interstage coupling circm s.

24. A system comprising a plural odd number of electric space discharge devices, interstage coupling circuits for coupling said devices in cascade connection, means for balancing waves from the output side of the last device against waves from the input side of the first device, and means for feeding waves resulting from said balancing action through said devices, each of said coupling circu1ts' forming with interelectrode capacities of the two devices which it couples, a network that has two input terminals, one connected to the cathodes of said two devices and the other connected to another electrode of each of said two devices, the impedance presented by said network to said terminals being approximately pure resistance impedance over a substantial frequency range.

In testimony whereof, I have signed my name to this specification this 13th day of September, 1929. v

i FRY NYQUIST.

a plate current supply path for said one tube,

a resistor included in said path and in said coupling means, and means for rendering the phase shift in waves transmitted by said one tube to the other tube substantially zero, said means comprising an inductance in series with said plate resistor.

22. In combination, a transfer path for waves of a considerable frequency range, a reactance included in said path, an impedance