US2914738A

US2914738A - Adjustable correcting networks

Info

Publication number: US2914738A
Application number: US559965A
Authority: US
Inventors: Oswald Jacques
Original assignee: Compagnie Industrielle des Telephones SA
Current assignee: Compagnie Industrielle des Telephones SA
Priority date: 1956-01-06
Filing date: 1956-01-18
Publication date: 1959-11-24
Anticipated expiration: 1976-11-24
Also published as: NL97865C; BE544310A; FR70150E; FR1116277A; DE1224370B; US2991433A; FR76851E; DE1200883B; FR76852E; CH331380A

Description

Nov. 24, 1959 J. OSWALD ADJUSTABLE CORRECTING NETWORKS Filed Jan. 18, 1956 Ra K6 R L R R R2 1? Fly. I 2

Ra R6 5 T 4 I R [Ni (N70 Arm/0996 NOV. 24, 1959 OSWALD 2,914,738

ADJUSTABLE CORRECTING NETWORKS Filed Jan. 18, 1956 4 Sheets-Sheet 2 F/']./0 I zwvavro $460055 GSWALD ArroAWEK Nov. 24, 1959 J. OSWALD 2,914,738

ADJUSTABLE CORRECTING NETWORKS Filed Jan. 18, 1956 4 Sheets-Sheet 3 I/v yaw-0 J/ICQUAS 05 MILD Afraid/E),

Nov. 24, 1959 J, OSWALD 2,914,738

ADJUSTABLE CORRECTING NETWORKS Filed Jan. 18, 1956 4 Sheets-Sheet 4 INVENTOR 77160055 05 W4 LD Br Q1 it St te Pawn: Q

The equalisation of the" attenuation curves of cable 7 sections with symmetrical or coaxial pairs is normally effected by means of well-known networks: equalisers, artificial lines, correcting networks, etc.

. However, it is 'very useful to have available; in .addition to the fixed correcting networks, adjustable correcting networks of which the attenuation curve can be modified as desired by a change ..of an impedance and even preferably of one or more resistances incorporated in the circuit diagram of the correcting network. Actually such networks make it possible to compensate thevariations of attenuation as a function of the temperature, as also 7 the residual distortions which may appear after a certain number of amplification sections.

As these networks must generally be connected between impedances which are comparable to pure resistances, constant image-impedances are imposed on them,

Patented Nov. 24,

Figure 8 shows a correcting network known in the art which can be compared to the one described in Figure 6;

' Figure 9 is a modified form of the correcting network describedin Figure 6;

Figure 10 is a diagram representing the operation of the correcting network according to Figure 9;

,; tion-is characterisedin this that it'cornprises a cut-all antimetric 4-terminalnetwork, composed of 'a lattice-type Figure 11 is a duplicate of Figure 3, and I Figure 12 represents the circuit diagram of network Q. appearing in Figure 11.

-The correcting network in accordance with the invennetwork of which twoopposite branches are reactive and inverse, the other branches being reduced to variable resistances of which the product remains equ'alto that of the reactances of the reactive branches.

At first, it is believed best to set forth the definition of certain terms as used herein in both the specification'and:

claims,

An antimetric quadripole is a quadripole having a mean characteristic impedance which is constant, i.e., a

quadripole whose product of image-impedances is constant A cut-all quadripole is a quadripole offering infinite impedance at all frequencies;consequently, a quadripole in which nocoupling exists between the input and output 1 terminals thereof, i.e., the'impedance as seen'from one* of its pairs of terminals is completely different from the impedance disposed between the terminals of the other pair.

A lattice-type cut-all antimetric filter" is, therefore,

'3; nothing elsebut a bridge in equilibrium, whose imped which, moreover, make it possible to put them in series undereasy conditions. I I a I H. W. Bode has described a well-known class of correcting networks in French Patent No. 820,760, filed on January 30, 1937, and its first certificate of addition No. 49,867, filed on September 21, 1938. o

Fig. 4 of the above mentioned certificate of addition represents a circuit diagram of a correcting network in a form which is particularly used, and which is reproduced'inattachedFigJ 1.

a resistance R, in the superior branch which is in shunt or parallel with both horizontal branches, the network Q and, V in the shunt branch a network Q 1 these networks being inverse in relation to the resistance R and respectively closed on variable resistances R and R so as to remain constantly inverse in relation to R. The shunted T therefore has the constant,characteristic im pedance R. r

The same is also true of Fig. 5 of-the:aforementioned French certificate of'addition, reproduced ongattached Fig. 2, if there be added to the foregoing diagram, in the superior branch, a resistance R in parallel with the in- V This" correcting networkconsists of i shunted T, comprising in each' of its horizontal branches put of Q and, in the shuntbranch, a resistance R in series with the input of Q under the condition: R R =R The features of the invention are. pointed out .in the following description and drawings, in which: f J l r Figures 1 and 2 represent circuit diagrams, known'in the art, of correcting networks as mentioned above;

Figure 3 represents acircuit diagram of a shunted T correcting network according to the invention;

Figure 4 shows the circuit diagram of network Q appearing in Figure 3;

' g 5 represents correctmg network L work of infinite attenuation carried outinthe form, shown a differential transformer;

Figure 6 shows a correcting network as in Figure 3 in which a.4-terminal network Q is inserted for correcting the slope;

Figure I is a'diagram representing the operationof I the correcting network according to Figure 6;

whereas thebranches ances of each of the groups of opposed branches form a constant product equal to the square of the characteristic impedance R of the cut-allysuch a filter is illustrated in its 'most general form'in Figure 12 of the drawingin and In the schematic view of a correcting network according to the present invention,pthe branches 1 at Z,

are constituted by adjustablefresistances R, and zap- L fit Z1 r are purely reactive in. the special case examined herein.

A shunted'T correcting networkaccording to the invention of which the diagram is shown in attached Fig. 3, has the advantage over the correcting network according to the aforementioned certificate of "addition'49867 or only using, instead of two networks Q and Q a single network Q in a circuit connected onthe one side in shunt with the resistance R and on the other via a resistance R to the primarywinding of a transformer of which the secondary winding is in the shunt branch of the T. The network Q is an antimetric four-terminal net- Z1 and the other branches being composed of the variable resistances R, and R}, which are so'selected that the product The superior branch of the correcting network: is. therefore constituted by placing in paralleli with R themput impedance W of Q, which is the input impedance with open circuit as well as the image-impedance by reason of the balanced mounting; the impedance seen from the other terminals is 7' If' oz i's'the: transformation ratio, the: shunted T will certainly be of constant characteristic impedance R if o R sin a RIRZ R02 which requires R1R2=Ro The-theory of thisnew type ofcorrecting network will be given. Then the attenuation formulae obtained will be compared with those of the Bode correcting networks.

It-is known that the shunted T-type four terminal network of-characteristic impedance R with a superior branch l and with-ashunt branch 7 1 2 r admits a transfer exponent =a+jb given by the formula: (11) As has been indicated,.Z is constituted here by placingin parallel R and the'input impedance. W given by:

o? W1 R,,+z V ItIis' convenient to introducethe reduced impedancesz' PL L n o 1 o 0- a Z2 1" -z -z R R0" R and to put: a (fl) z-=coth 1 v I fail (5) V p--:-n+ Y which enables us to write g -ar 7 1W T 1- :pe'fs and consequently v J t W 1 1' 1 e'" 1 V w +r 71+]. 1 T '1e V 11+]. 7 and by carrying into (1) a v t -21 8 a 1 A l n+ M This-formulashows that, for =0, that is to' say =R the attenuation of the correcting network A is constant at all frequencies and given by (12) Formula 11 is simplified by choosing r by:

r =ei 0 hence it follows that We then get from 10' and 11' eed 0a 9 (1.4) tanh 2 .pa tanh 2 Formula 14 is strict, but its use a little laborious. It is possible to substitute for it a more easily handled approximate formula:

v For fairly small differences 0-0 it is possible to use the classic development of the logarithm:

restricting ourselves to the second term, the third being negligible inthe'existing conditions, which gives the sufficiently approximate formula:

(16) 9:a +2pe-= tanh By putting I =A+ 'B and by separating the terms relating to the attenuation and the phase-shift, we get:

7 a2a +2 r tanh cos 2B It therefore be seen that the: attenuation of the correcting. network comprises in addition to" the.fixed portion a ,;a portion proportional to At the; frequencies for which E is anoddmultiple of cos' ZB' 1 is ze rci 'and' a equals a whatever the value: given to the parameter of adjustment p: we have pivot points ofitheattenuationcurvea V r l v Choo 11g for Za pure reactance; jX'(w), is'l reduced The pivot points therefore correspond tothe values if of the reduced reactance; there may be any number of them, and the number is reduced to one for an inductance or' a'capacity. We have:

(A)- agil -F2 tanh cos 2B' The maximum differences occur at the frequencies for which:

that'is, for the zeros and the poles: of Z.

It may be pointed out that it is sufficient toreverse. the

' and one may,

b? again? 1 iandIR' is equal to input and output: terminals of the 4-t e'rminal network' inorder to reverse this quadripoleas seen from the shunt branch R; is designated Wil; The symmetrical lattice equivalent tothe I differential has as branches; m

tRz-lr 'zh 23:1; W4 (notation for placing in parallel). 1

"That equivalent to the T ha's' as branches] 1 new v we and, inorderlt hat: W I r r f' t. it necessary to take l "Q may be dedu'cted froin' the' quadripp'le '-Q by the reduction to one-quarter of the impedafices of the branches.

The character impedance R 55, thereof is equal to therefore,v vrite: w I

n the other hand, R' is e'qual to A II 2R, R that is tinita ns i c nsider o E u i s 12 and to 1'4 cosh a ,The Bode correcting networks give transfer exponents defined by the Formula 16 in which. a and p have the same ineaning (reference attenuation and regulating pa- Qrarneter which in thiscase isthe coefiicient of disadaptation to the input of Q; arid Q but in which I is the transfer exponent (efiective'or on images) of Q and Q acteristic.

.P i If the networks Q aha arephase shifters, and if Z=jX be the common value of the reactance of two opposite branches, we have: v

B .X 7 t -w whence c0tvE .'R n z 3 j, i and we also have the formula (A); (A ame,:t'a n gfosfze Theipiyots, howeven correspond the frequencies for which the modulus of the reduced reactance x is equal to In thiscase, therefore, there are at least twopivots, and

always an even number. a n I It will therefore be seenthat the new correctingnete works permit more freedom in the choice of the char- As has already been pointed out, the reversal of the input -4-t erminals of thequadripole Q makes it. possible to reduce the number of resistances 'by a simple arrange:

value p=1.

ment, while this is not the case with the Bode' correcting networks. q w e 3 i H The following examples of use will make it possible to forrna more complete judgment of the advantages of the correcting networks according to therinvent on par: ticularly with regard to savingin reacitveelernents. H First of all we will take the example of ;a.slope-correcting network, of irnpedance equal to ohms, admitting a pivot v frequencyj of 68 l c./s. with an attenuation f 1 12 to 68 kc./s. band: 1 n Fig. 6 represents the simplest correctinglnetwork.;set up in accordance with the method described. a

The reducedresistance I: a 1

of 0.8V nepers, the correcting network being for use inv a is equal to This is theeratio .oft thetransformer R is. equal to The antimetric cut-all 4-t e1 minal network 'Q, "will allow the frequency of 68 kc./s., which fixes the valuerif'the capacity: C=8840 pF and of the inductance of the opposite branch: L=CR =625 uH.

The approximate Formula A here gives: 22 earn-{ 0.76 ,0 cos 2B B being determined Q I (23) cot B Q g The exact formula has been applied for the maximum The correction curves have been represented in lines in Fig. 7 for the values p f0.5, p- ;1. J We will determine the corresponding Bode correcting network. v y l 1 Let us choose for Q an elementary phase shifter composed of a symmetrical inductance-capacity lattice.

mitting pivots at the frequencies 12' and The application of the formulae previously given makes it possible to calculate the attenuation of the correcting network, the curves corresponding to piOj, ;':1 are represented in discontinuous lines in Fig. 7;

The determination Of'the elements is as follows:

The superior resistance is of 591 ohms, as in the correcting network according to the present invention; the

resistance in the shunt branch of the T is equal to correcting network comprises 4 inductances and'4 capaci'ties;

The-comparison of the attenuation curves shows their slight difference; neverthelessthose of the correcting net'- work accordingf to the invention are morerectilinear, for the-lowvaluesf of p: i i

Again we will take theexample of a curvature cor rcctinganetworkof the'sat'ne impedance 150-ohm's;ad'-

an attenuation a of 0.42 N. 1 a

We have here? V tanh =tanh 021%(1'207 V The existence'of the two, pivots requires, as the simanda-ratio equal to plest solution, the use of an auxiliarynetwork of which the reactive branches are composed one of a resonant circuit (L, C), and the other of an antiresonant circuit V The elements of the resonant circuit are immediately 1 determined" by: the value ]R}, of'its'. impedance on the pivots: i r r V 9 gives 'the'diagram of the correcting network,

Big. 7 10; the .attenuationcurves for the samevalues of p as before; Elle have therefore seen, that; the methodcon- ,sisting, of'using 'cjut all four-terminal networks in. the correcting network permits of advantageousresults.

, In the1 foregoing aparticular choice of fcut all? networksibeen i dicated; we have assumed two opposite reactiverbranches and two composed of resistances, of which the conjugate variation makes it possible to obtain adjustable variation attenuationsz- It should be note'd'ithat, generally speaking, the use of cut-all four-terminal networks, with reverse opposite. branches, withthe same ratio of inversion may be of interest inadjustable or fixed correcting networks.

-Fig. 11 represents a.correcting network of this kind in the form of a shunted T, R may be infinite and R 7 zero. The general plan of the four-terminal network Q is. given in Fig. 12: this is a lattice with opposite reversed branches, Z

, Z1 and: 2," 7 I V 0 Z which branches may be purely reactive, purely resistant or comprise an assembly of reactive and resistant elements. The property of a four-terminal network of this kind to be capable of being substituted for the two fourterminal networks of the Bode type diagrams, by the use of its open-circuit impedances, seen from each pair of terminals, may lead to a saving in the elements used in the setting up of the correcting network.

What is claimedis: I 7

l. A correcting network with adjustable attenuation and having a constant characteristic. impedance equal to R, formed by a symmetrical bridged-T circuit comprising horizontal branches each-having a resistance equal to R, a superior branch in parallel with said horizontal branches having a resistance equal to R an antimetric cut-all lattice-type quadripole with one pair of the terminals thereof connected in parallel with said superior branch and having a characteristic impedance R whereby R is related to R and R; by the equation i RR:12.' t R0 R i 1 a. transformer havinga primary anda secondary winding said secondary winding forming the shunt branch connected to the midpoint of the; horizontal branches of said network, a resistance R equal to said primary winding being connected across a= circuit including said resistance R and the other pair of terminals of said quadripole, said cut-all lattice type quadripole including two opposite branches with adjustable resistances of which the product remains equal to R and two other opposite branches having fixed impedances whose product is equalto R 2. A correcting network according to claim 1, in which said fixed impedances of said cut-all quadripole are reactances.

3. A correcting network" with adjustable attenuation and having a constant characteristic impedance equal to R, formed by a dilferential network equivalent to the T-type comprising a hybrid coil with two identical primary windings and a secondary winding, said two identical 7 two opposite branches with adjustable resistances whose product remains equal to R and two further opposite branches with fixed impedances whose product is equal to R a shunt branch in said difierential network including a resistance R connected across the other pair of terminals of said quadripole, the values of R and R,

being equal to I RR and

respectively, and the ratio of the number of turns of said secondary winding to the number of turns of each of said primary windings being equal to zontal and shunt branches, a transformer having primary and secondary windings, one of said windings being symmetrically connected in one of said horizontal and shunt branches of said network, impedance means symmetrical 1y forming the other of said branches, an antimetric cutall four terminal lattice-type quadripole having infinite attenuation and with a constant mean characteristic impedance, means including a resistance for connecting one pair of terminals of said quadripole to the other winding of said transformer, the other pair of terminals of said quadripole being connected across said impedance means, said quadripole having two branches including adjustable resistances whose product remains equal to the square of the characteristic impedance of said quadripole and two further branches including impedances whose product is also equal to the square of the characteristic impedance of said quadripole.

References (Iited in the file of this patent UNITED STATES PATENTS 1,732,311 Nyquist Oct. 22, 1929 2,044,047 Bobis June 16, 1936 2,070,668 Lundry Feb. 16, 1937