US2380923A - Negative feedback amplifier - Google Patents

Description

Aug. 7,1945. J. R. CANNON 2,380,923

NEGATIVE FEEDBACK AMPLIFIER Filed Nov. 16, 1945 2 Sheets-Sheet 1 I11 V67? al; ames T. Camrzon Aug.'7, l945.- J. R. CANNON NEGATIVE FEEDBACK AMPLIFIER Filed Nov. 16, 1943 2 Sheets-Sheet 2 Patented Aug. 7; i945 I Nsua'nvs James Rowan FEEDBACK surmusa Cannon. Lives-poo signer to Automatic Telephone ted, lives-pool, Englan a comp! B of Great Britain mm time 0, ms, suin'na siosss I p In Great Britain se -m .e. in:

" 1 can. (Cl. 110-111) The present inventionrelates to negative foodback amplifiers and is more particularly concerned with push-pull amplifiers of the negative performance multi-stsge amplinsn' of the abovedescribed type.

- According to the invention, in a negative feedback amplifier ofthe multi ltase push-pull ype.

toxreduoe harmonic distortion, the control for at least one stage are derived from a resistance-common to the cathodes valves of the stage to reduce spurious oscillation and all the valves type which. are rendered the following description of one'method of carryit' into elect. reference being had to the accompanying drawings comprising Figs. 1 and '3 which when placed side by sideshow circuits'oi a two-stage push -pull negative feedback amplifier 7 suitable for use in a multi-channel carrier freoutput to line. this is eifected we multi -windin harmonic distortion relates principally to the I ooresoi' magnetic material having-suitable airamployed areof the v better understood from tive high tension lead connecting with the screen grids of valves VA and VB via resistance YG. while it also connects via this resistance and then via choke Ll which is centre-tapped, to the anodes of these valves. As regards the valves VC and VD, HT positive extends to the screen grids via the centre-tapped resistance VP and to the v anodes via the centre-tapped primary winding of the output transformer 0'1; Condenser QA. located between the two stages. serves simply for decoupling purposes. iv i Consideringnow the operation of the amplifier, carrier frequency speech or other signals received at the input terminals ll, II, are applied to potentiometer PA which is graduated in decibel (db) steps and-thence the applied voltage is extended via resistance YA to the primary winding of the input transformer 1'1. Attransformer IT it divides between the two secondary windings and is applied to the; control grids of valves, VA and VB via the grid stopper resistances Y8 and Y0. the standing bias potential onthe controlgrids being derived from resistance YF which connects the cathodes of valves VA and VB with the high tension negative lead;- The applied signal is amplified by valves VA andVB and is applied via the chok-capacity coupling including choke LI and condensers Q8 and QC to the control grids oi the valves VC and VD oi the second stage. resistances YK and YB serving as grid stoppers, and Y8 and YJ as leak resistances.

On these two valves the standing grid bias poten-- tial is obtained by means of resistances YM and YN which are individually associated with the respective cathodes and which are"by-pas sed respectively by condensers QE and 616 to prevent T unwanted negative feedback voltages from being second harmonic whiehoan be substan ial y bale ,5 anced out in a push-pull circuit; The circuit arrangelnents are'moreover suchthat these valves need not be in'themselves perfectly matched as regardsm the cars ture oi their charac 'teristic so: the use or simplification uie valve filament circuits have'been'omitted; as use have the valve testing circuits and supervisionand'alarm equipm nt which isms-ll: provided on such amplihi's'h tension voltage supply is conwinding is communicated to the line winding and from there to the line output terminals ll and it. The winding is used for voltage feedbackpurposes and the voltage appearing across this winding to thewouter terof the valves VA and VB in the first push-pull stage where it functions in known manner partially to oppose the original input.

In order to obtain the desired degree of performance from the amplifier, it is necessary to" prevent spurious oscillations being set up, while it is also necessary in order to avoid inter-chan- .nel interference to minimise harmonic distortion.

Spurious oscillation in a negative feedback '2 assaaas substantially equal and opposite in phase and therefore no feedback voltage is developed. At

' the second push-pull stage, the phase shift cited;

amplifier may occur at a very high and/or a very low frequency, and even though the frequency concerned may lie outside the normal working range of the amplifier, its presence will seriously change the operating conditions of the valves and hence will affect the normal working of the amplifier. Such oscillation is usually due to the feedback having become positive instead of negative, for one thing as a result of the phase shift through the amplifier being different for different frequencies due to variation of impedances of inter-stage coupling transformers, etc. Thus the feedback at high and low frequencies may become 180 out of phase with that at the middle frequencies of the working frequency range, in other words the feedback at extremes of frequency becomes positive instead of nega-.

tive. If the loop gain :13. where u and B are the complex transfer constants of the amplifier and feedback circuits respectively, becomes unity at these extreme frequencies where the angle of a3 is zero, oscillation will occur. The exact criterion'for stability is described by Nyquist in the "Bell System Technical Journal 1932, page Furthermore, the conventional method of obtaining automatic grid bias potential ina pushpull amplifier is by means of a voltage'dropping resistance and associated by-pass condenser in the circuit of each valve cathode as is done in the high level second stage of the amplifier beingdescribed, the condensers serving to prevent the development of unwanted feedback voltages across the resistances. At high frequencies the condensers provide a shunt path of negligible impedance for, the A. 0. component of voltage across the biassing resistances so that there is no feedback of signal frequency to the associated valve grids and the phase of the anode currents is therefore not affected by the .presence of the condensers. At lower frequencies, however, when the reactance of the condensers is no longer negligibly small compared with the resistances, the phase of the anode currents is given a, shift in the leading sense by the condensers. v

-In the case of the first push-pull stage where this phase shift is additive with that produced by the high series impedance of the inter-stage coupling condensers and the low shunt reactance of transformers and chokes which also occurs at low frequency, it is found that by deriving the automatic grid. bias for the valves VA and VB of this stage from common resistance YI" the phase shift is minimised with a resultant increased stability. resistance Y1", there is no necessity toinclude a by-pass condenser. since, with the push-pull arrangement of valves the twp A. 0. components As regards the common bias is small compared with that of the first, so that it has not been considered necessary to employ the common bias resistance as in the first stage, but it is appreciated that it is in no sense essential to use a common bias resistor in the first stage only. The use of the conventional automatic bias arrangements for. the second stage proves to be of convenience as will now be described. p

In the low level push-pull stage, mismatch between the curvatures of the characteristic curves of the two valves concerned is not particularly serious from the point of view of harmonic distortion since the applied signal level and resultant grid swing is small. In the highlevel stage, where large grld swings are concerned, the mismatch question is much more serious and here the effect of close limit matched valves is obtained by means of the variable resistance YO in the cathode .circuit of valve VD, by means of which the working points on the characteristic curves of the two valves can be located on similar curvatures.

Considering further the reduction of harmonic distortion in a negative feedback amplifier, the use of current feedback ordinarily obtains this result. When, however, the output of the amphher is takenvia an output transformer, as is usual in carrier telephone 'circuits 'where' a matched connection between the amplifier and line has to be made, it is necessary to use only voltage feedback, since 1: current feedback is usedeither alone or as well as voltage feedback, harmonic production may well be increased. The reason for this is that voltage feedback tends to maintain the output voltage sinusoidal which is the condition required but current feedback tends to maintain a sinusoidal current in the output transformer. Owing to the non-linear nature of the relationship between current and magnetic fiux in the transformer iron circuit,

.this results in a, non-sinusoidal output voltage with resultant harmonic distortion. On one test of an amplifier employing both voltage and current feedback, removal of the current feedback was found to improve the harmonic content by 10-15 db, and at the same time the gain was increased. by about 2 db. I

A further point as regards harmonic distortion is that in a push-pull amplifier using close limit matched valves and an accurately balanced circuit, all even harmonics are reduced'to an extremely low level, and, owing to the careful balancing'oi' valves it is usually assumed to be willcient to design the iron circuits of push pull' chokes and transformers to carry a small percentage of the steady magnetising anode current of each valve, this catering for differential ageing of'the valves,.etc. It was found, however, that it was possible for substantial uni-directional current pulses applied to or produced in such a pushpull amplifier to cause unequal currents to flow momentarily in the two sides of the amplifier. This caused a small amount of permanent D. C. magnetisation of the iron cores to take place with consequent deterioration of the second harmonic distortion. which while of no account in a-single channel amplifier, would be serious in a multiof the. anode currents flowin therethrough are been made of the gapped type.

In an actual amplifier where the cathode bias resistances of the high level push-pull stage valves VC and VD are replaced by the windings of supervisory and alarm relays, if one of the relays was momentarily operated by hand the resultant m0- mentary voltage induced in the winding was sumcient to cause a permanent deterioration of the second harmonic content of about 20 db dueto D. C. magnetisation ofthe output transformer. The substitution of another transformer having a butt joint air-gap was found to eliminate the e t e re y and no deterioration in harmonic content occurred even when one of the valves was removed and replaced while the amplifier was in operation.

As regards other details oi'the circuits which have not been described above, the resistance YA pedance matching when looking in from the line and when lookingout from the valves in the high level push-pull stage. Condenser QH across the output terminals i2 and it improves the stability of the amplifier and flattens the impedance and frequency response curves.

Tests which have been made on the amplifier described show that complete stability 'is obtainable even under conditions such as short-circuited input and open-circuited output. The tests show further that there is notendency to self-oscillation when an input and/or output valve was removed, either under normal or abnormal terminal conditions.

What I claim as new and desire to secure by Letters Patent is:

connected in series with the primary winding of the input transformer YD serves to further the margin against instability due to spurious oscillations by reducing the phas shift of the feedback i voltage which occurs in the input transformer at low settings of the gain control potentiometer PA when the primary winding would normally be terminated, as far as locking-in to line from the amplifier is concerned, by a low resistance. The

resistance YA-has a negligible effect on the am-- plifler gain. As regards resistance YQ in circuit with the output windings of transformer OT, this serves for loading matching purposes and is arranged in conjunction with a suitable turns ratio A negative feedback amplifier of the two-stage;

push-pull type employing chok capacity coupling in which the cores or the inter-stage coupling choke and of the output transformer are provided with suitable air-gaps to reduce harmonic .distortion, the control grid bias potentials for the on the transformer so as to give the proper imfirst stage valves are derived from a resistance common to the cathodes of both valves to reduce spurious oscillation and all the valves employed are of the beam power type which are rendered satisfactorily interchangeable in the secondstage by the provision of separate resistances in the cathode circuits of the two valves with adjusting means for at least one of said resistances.

JAMES Rowan CANNON.

Images