US3004703A

US3004703A - Calculating machine with polarized relays

Info

Publication number: US3004703A
Application number: US445069A
Authority: US
Inventors: Hoppe Walter
Original assignee: El Re Ma S A Per Io Sfruttamen
Current assignee: El Re Ma S A Per Io Sfruttamen; El-Re-Ma S A Per Io Sfruttamento Di Brevetti
Priority date: 1953-07-31
Filing date: 1954-07-22
Publication date: 1961-10-17
Anticipated expiration: 1978-10-17
Also published as: DE1074886B; GB780953A; FR1111050A; NL104334C; GB780954A; US3004704A; CH316201A; DE1131036B; NL189405B; NL188401B; CH316200A

Description

Oct. 17, 1961 w. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS l6 Sheets-Sheet 2 Filed July 22, 1954 A illlllll nv AVIUENTQE Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 2 n I I Wm M Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 3 3411 56 I! ]II[ 58 1 59 IMF "1 38 H @AAAgA El 51 Oct. 17, 1961 w. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS l6 Sheets-Sheet 4 Filed July 22, 1954 #mswne 10% By a Qty-MY Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 5 wow Oct. 17, 1961 w. HOPPE 3,00

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 6 Oct. 17, 1961 w. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 7 MVE/Yrog M W 31107 K. )4

Oct. 17, 1961 w. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 8 ##orney Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 9 Oct. 17, 1961 w. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 10 2 2 r P. .Dl l um rf w R 1 W a A D fi w A =1 v 2 d C I. I i l A W E W. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS Oct. 17, 1 961 16 Sheets-Sheet 11 Filed July 22, 1954 Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 12 roe 580 (6 $06 VVENTO k 19% 1f" M Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 13 /NVENTOIE flat. M

Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 l6 Sheets-Sheet 14 mus/Wok Oct. 17, 1961 w. HOPPE CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 15 Oct. 17, 1961 w. HOPPE 3,004,703

CALCULATING MACHINE WITH POLARIZED RELAYS Filed July 22, 1954 16 Sheets-Sheet 16 4% -/M/E-Tp w w BY M Walter Hoppe,

United States Patent CALCULATING MACHINE WITH POLARIZED RELAYS Bern, Switzerland, assignor to El-Re-Ma S.A. per lo sfruttamento di brevetti, Lugano, Switzerland, a corporation of Switzerland Filed July 22, 1954, Ser. No. 445,069 Claims priority, application Switzerland July 31, 1953 25 Claims. (Cl. 235-159) This invention relates to electrical calculating machines.

More particularlythe invention concerns an electrical calculating machine controlled by current impulses, comprising at least three groups of contacts for introducing into the machine the numbers involved in an operation and for performing that operation, and a group of controls determining the kind of operation to be performed.

Such calculating machines are-already known, but they are generally too cumbersome and too heavy to be portable, because of the 'considerable number of relaysthey comprise. 1 i

It is anobjectof the calculating machine with which the invention deals to correct this drawback; it 'diifers from previously-known machines in that the contacts of at least one of these groups are controlled by polarized relays with at least three positions, one of these positions being the rest position, these relays making it possible to perform certain operations by reversing the direction of the current impulse acting on the said group. This characteristic makes'it possible to save contactmakers and switching elements, and to simplify the calculation diagrams and-their control mechanisms, as compared with machines hitherto known.

It is desirable to use relays with windings for electromagnetically polarizing their armatures; this makes it possible to choose the direction of polarization at will so as to perform certain control procedures.

It is preferable to choose the value of the polarization windings and the air-gaps of the said elements in such a waythat the armatures when attracted are kept in the attracted position by the polarization action alone, even when the main winding is no longer excited,'until the polarization is cut off. The elements used are thus selflocking, without providing special electrical or mechanicallocking devices. i

It is well known that the possibility of locking contactmaking elements is of basic importance for the construction of machines of the above type. Locking of the said elements in several positions by the polarization windings has still other advantages, as it takes place without inertia, since no other mechanical part, neither a bolt nor anything else, is actuated. Actually, movement during calculating operations is confined to the movements of light. armatures moving a'fraction of a millimeter; this allows great speed of calculation to be attained.

The following description indicates, by way of example and with'reference to the subjoined drawings, several embodiments of a calculating machine according to the invention.

FIGS. 1 to 9a represent various relays and groups of switching elements that can be used in the said machine.

FIG. 10 schematically illustrates a calculation circuit that makes it possible to perform the addition of the numbers defined in the A and B groups, the result of which is determined by the contacts of the R group.

FIG. 11 shows schematically a transfer circuit that makes it possible to transfer the result in the R group to the B group.- v l FIG. 12 illustrates the principle of converting a number into its nines complement.

FIG. 13 is a diagram of the current impulses whose purpose it is to actuate relays.

FIG. 14 schematically represents a rotary commutator for producing the impulses of FIG. 13.

FIG. 15 shows, very schematically, how the calculating and transfer circuits are fed by an impulse generator Ig.

FIG. 16 schematically represents the principal circuits of the machine.

FIGS. 17 to 20 illustrate the principle of ordering the operations.

FIG. 21 shows another impulse diagram that can be used to actuate the relays of the various circuits.

FIGS. 22 and 23 represent an impulse generator for producing the impulses of FIG. 21.

FIGS. 24, 25 and 26 represent the calculation and transfer circuits of another embodiment of the machine, in which the representation of the numbers is made on the decimal system.

FIGS. 27, 28, 29, and 30 refer to a variant of the calculation and transfer circuits in which the representation of the numbers is made on the odd-even system.

The relay shown in FIG. 1 comprises a magnetic core 1 bearing a control winding 5, and two

armatures

2 and 2a that are urged against

blocks

6 and 6a by springs l and 7a. Input conductors 8 and 8a in the form of wires or flexible strips are connected to these armatures, while a third input conductor 9 is connected to the core. The poles of

armatures

2 and 2a, which are U-shaped, plunge into

polarization coils

10, 11 and 10a, 11a respectively and are polarized by the latter. The ampere-turns of these polarization coils are so chosen that when the effect of a polarization coil is opposed to that of the control coil, the resultant magnetic field is not sufficient to cause the attraction of an armature subjected to this field.

When a current is sent to control coil 5, for example, so that core 1 has its south magnetic pole up and its north magnetic pole down, armature 2 is attracted while armature 2a is not for the upper end of armature 2 has been polarized to show a north magnetic pole, while the upper end of armature 2a has been polarized to show a south magnetic pole at its upper end. If the direction of thecurrent is reversed, either in

polarization coils

10, 11, 10a and 11a, or in main coil 5, armature 2a is attracted. The armatures make theelectric contact at their two ends, in two points, which increases the certainty of this contacts functioning. The contact points of the armatures are coated with a non-magnetic metal which is a good conductor, such as silver or copper, in order to assure a good electrical contact and to diminish the remanence of the magnetic circuit. This relay forms a switching element that can have the following three positions:

0:

armatures

2 and 2a are not attracted (no current in main coil). Input conductor 9 is not connected to 8 and 8a.

I: armature 2 is attracted, 9 is connected to 8.

II: armature 2a is attracted, 9 is connected to 8a.

If the intensity of the current in the windings, the maximum gap between core 1 and

armatures

2 and 2a, and the thickness of the non-magnetic metal coating on the contact points are correctly determined, the armature that has been attracted, 2 or 2a as the case may be, can be maintained in its position, after the excitation of coil '5 has been broken, by the action of polarization coils 10 and 11, or 10a and 11a, respectively. When the current is broken in the polarization coils, the attracted armatures are brought back to rest position by

springs

7 and 7a.

FIG. 2 represents a group made up of five relays of the same kind as those shown in FIG. 1, except that in them each control coil 5 surrounds two parallel cores 1 that are magnetically and electrically insulated from each other. Each of these cores cooperates with two

armatures

2 and 2a.

parts 32 of insulating material. to support. these. elements.

rounding the lower end' of the. armatures 2 of the five elements and another common coil 10' surrounding the 7 upper. end of these armatures.

The relay shown in FIG. 3 has a core 1 surrounded by a control coil 5. Polarization coils 10 and 10a are mounted on: studs 12. and. 12a, which are. separated from the core by insulating

layers

13 and 13a and bear'parts 14 and 14a, on which armature/

s

2 and 2a are fastened by meansof fiat springs; 15: and 15a. A part 16 made of insulating material serves as a stop. for" the armatures and makes it possible to set the maximum air-gap between the latter and the core.

Excitation of

polarizationcoils

10 and 10a polarizes the armatures, and when the current flows through control coil 5,

armature

2 or 2a is attracted, according to the direction of the polarization.

It is obvious that several relays. like the. one in FIG. 3

could be combined to make an assembly of the same sort as the one in FIG. 2.

The combination of magnetic circuits and electrical contacts, as shown in the preceding illustrations, has ad.- vantages for the construction of the relays, but still is not indispensable.

FIGS. 4 and illustrate a relay of the same. typeas-the one. in FIG. 3,. but with an added feature, a means for showing the position of an armature. The core 1 of this relay is longer and a lever 27 is attached to its end on anaxis 28 around which it can rotate. A spring urges this lever against a movable stop 30 that can be shifted in the direction of arrow 31. When this stop 30 is shifted in this direction, spring 29 causes lever 27 to rotate; the lower end of lever 27- then comes between armature 2 and core 1. If armature 2 is in attractedposition, that is, in contact with core 1, when stop 30 is shifted,'the lower end of lever 27 strikes against armaturev 2. and this. lever remains in the position shown in FIG. 4, instead of being able to put itself in alignment with core 1. The upper end. of the. said lever thus describes a path that is much longer than the displacement of an armature; this. permits the use of a simple apparatus for mechanically exploring therelayswhose armatures are attracted, without requiringv very closemachining tolerances.

The relays described above suflice for the construction of'acalculatingmachine according to the invention. However,, in the embodiment described below, use. will also be made of multiple-contact mechanical wipers; an example of. their construction is shown in FIGS. 6 and7.

Ingeneral, thesewipers comprise agroup of movable points moving across fixed points and, because of the friction of'these-points against each other due to the pressure required toinsure a good electrical contact, a considerable force is required for shifting the movable points. To avoid this drawback, it is preferable tohave wipers whose points are not in contact during their respective movements, the contact pressure between. fixed and movable points occurring only after the latter have moved into position. Wipers of this-kind, in which the movable and immovable. points: are made of a magnetic material: and enter into magnetic circuits that can be excited by means of a winding, are described in detail in application No. 356,577, new Pat. No. 2,854,541.

On a sliding par-t 70- of insulating material (FIG. 6), whichis attached to rod-

s

71 and 72,, there arearmatures 73, set in channels in which they can make a small move-' ment. This; movement is: limited on: one side. byrods-74 and 75 and on: the other by magnetic cores 76; The

latter are insulated' from each other by insulating'layers 77 and. are surrounded by a winding 78). 80 longer; the

4 latter is not excited, part 70, bearing armatures 73, can be movedvirtually withoutv friction, and, depending on the number of armatures 73 and cores 76, can assume a certain number of positions for which each armature 73 is opposite a core 76.

When coil 78 is energized. for one of these positions, armatures: 73' areattracted'to cores 76'- and. exert a considerable pressure onthem at twopoints of contact 79 and 80.. Since these cores, '76 constitute the; fixed points and the armatures the movable points, a stable and certain electrical contact. is obtained inthis way, because of this contact at two points. It is, advisable to. cover these cores 76 and thearmatures with a. layer of precious metal in order to improve the electrical contact.

FIG. .8, is a view of a variant of a relay according to FIG. 3, in which, however, polarization windings 10- and 10av have been placed close. to thefree ends: of armatures .2 and 2a;

FIGS. 9' and 9a represent an. assembly made up of. five relays likethe-onein FIG. 8, inasimilar way to- FIG. 2. In this case, however, the polarization coils have been so arranged that every relay can assume five positions 0, I, II, III, IV, the. 0 position being the. rest position, in which no electrical. contact is; made. As. is seen from- FIG'. 9a, connecting the polarization coils in a suitable manner makes it possible to polarize threearmatures of a group of four in one sense and the fourth in theother sense, so. that by; sending through a control current with a suitable polarity we can selectively bring about the attraction of. a single armature out of the group. of. four.

The five relays of this. set eachhave a

core

56, 57, 58, 59 and 60. passing through mainwindings34, 35, 36, 37 and 33 respectively. Each core. is designed to cooperate with. four armatures capable of entering into contact with it.. Core 57' cooperates with

armatures

33a,,33b, 33c, 33d, which arepolarized by four

coils

42, 42a, 43 and 43a. If polarization coils 41- to 46 as well as. 41a, 43a, 45a are connected inthesarne direction, while

coils

42a, 44a, 46a are. connected in the opposite direction, and a current is sent through control coil 35 only, its direction being so chosen that only those armatures that are polarized by coils 42a, 4411,4611 are attracted, armature 33bwill be attracted: when. the control coilisexcited. By varying the connections and. the=current direction. in the polarization coils, the. attractiomas: desired,. of any of the four armatures 33a, 3312, 330, 3311. can be obtained.

Electrical calculation portion FIG. 10 schematically represents a calculating circuit that: makes. itpossible' to: perform: the-addition of the numbers; defined inthe A. and B groups; the result of which is defined-by the contacts: of theR group.

FIG- l1 shows schematically a. transfer: circuit that makes itpossibleztm transfer theresult in the R- group to the, B? group. l2..illustrates the'princiyglle of oonvertinga number into: its. nines complement;-

Thecalculating machinedescribed below comprises three groups of contacts A, B and R- The. A group is made. up of multiple+contacts wipers such: as in. FIG. 6, while the contacts of'the B and: groups: are controlled by polarized relays with at least three positions such. as in FIGS. 1, 3 and 8. The contacts of each group are capable ofdefining a: number'by'their'closed positions. The wiper andrelay. contacts ofthe respective. A- and B- groups areconnected. to the control.- w-indings: of. the relays: of the R. guoup in such. a way as to cause the. closure of. those contacts. of. the R. group thatdefine. the number correspondingto the addition of'the number defined by thecentactsof the A and B groups. The contacts of the R group are connected to the control windings 5 of-the relays of the B group in such a way as to cause the closure of those contacts of. the, B group that define a number that will be a function of the number defined in the R group, that is, the number itself or its nines complement.

The operation of the various calculating machines that will be described hereinafter has been so devised that the positions of the contacts of the A group remain the same throughout a calculation, as is indicated in detail in application No. 445,070. I v

In principle, a calculation consists of a series of twophase elementary operations. Each two-phase elementary operation itself is made up of a calculation phase and a transfer phase, the calculation phase being that in which the numbers defined in the A and B groups, whose result is defined in the R group, ar e added, while the transfer phase is that during which the result defined in the R group is transferred to the B group.

Calculation and transfer phases In order to make the diagrams clearer and get shorter connections, the contacts and control coilsorf the relays have been separated in the drawings, 'as is customary practice in telecommunication diagrams. It is to benoted particularly that the contacts and coils that are connected to each other never belong to the same group of relays. The circuits that make it possible to perform a calculating phase will henceforth be designated as R circuit, while those that make the transfer phases possible will be designated as B circuit. The control coils of the relays of the R group will be designated, for the purposes of the description, by Ri, where the index i represents the digit defined by the relay in question. When a relay can define several digits, one of them is chosen for the index i. A relay generally has two and sometimes three control coils, which are then designated as lRi, 2R1, etc. The contacts of the relays of the R group are designated in the same manner, but with an r instead of an R for their I position and an r* for their II position, these positions having been defined in connection with FIG. 1. One of the directions of the windings of the control coils has not been explicitly designated on the drawing, while the opposite direction is indicated by an arrowparallel to the symbol of the coil represented. In the diagrams reference signs are indicated for only one denomination order, as the signs for the other denomination orders are subject to the same rules. When the description must indicate that the relay in question is in a given denomination row, an index k is added (Rik, rik), which refers to the number of the denomination row. The designation 2R23, for example, refers to the second control winding of a relay of the R group, that defines the figure 2 in the third denomination row. 2R2, onthe contrary, refers to the second control coil of the relay defining the figure 2, without specifying the denomination row. The designation of the index k is omitted-for the reference signs of the drawing in order to make it clearer, the indication of the denomination rows being given generally by a brace for each place. This applies equally for the designation of contacts. In addition, when arelay has only one contact or one winding, the reference will be simplified by omitting the first figure. A single contact, for example, will be designated by ri instead of lri.

The control coils and contacts of the B. group are designated in the same manner, except that they have the letters B and b instead of the letters R and r. The relays of the B group thus have contacts lbi, Zbi and lbi 2bi* respectively for each of the positions I and II.

In the calculating machine, whose calculation and transfer circuits are shown in FIGS. 10 and 11, the figures are defined within each decimal place by the closure of contacts on the biquinary system. In other Words, seven contacts must be available to define a figure ranging from to 9. Five of these contacts enable us to define the figures ranging from 0 to 4, while the two other contacts are provided to determine whether a zero or a five should be added to the figure determined by one of the first five contacts. Thus in the R group, these first five contacts and the'windings of the relays that control them, will bear .an indexfroin 0 to 4, while the indices u and v are provided for the other two contacts. The figures of a denomination row will thus be defined, on the'biquinary code, by the closure of contacts whose indioes are given in the table below:

7 Di it index Hence, a number is always determined by the simultaneous closure of two contacts for each denomination row, both in the contacts of the A group and in those of the B and R groups, as described in application Serial No. 445,119.

The calculation circuit represented in FIG. 10 is that 'ofa simplified calculating machine with six denomination rows in all. In this diagram, asin that of FIG. 11, only the circuits of the relay control coils have been shown, the polarization circuits being omitted. The elements belonging to each denomination row are indicated by a brace numbered from 1 to 6. The units denomination row has number one, the tens place number 2', etc. To simplify the diagrams, the reference signs have not been indicated except for one or two of the denomination rows.

FIG. 12 illustrates schematically the contacts of the B group that are needed to define on the biquinary system the various digits of a denomination row, that is, the digits from zero to nine. These contacts are controlled by four relays with one, two or three control windings and polarization windings, the latter not being shown.

The first relay has a control winding B2 and four contacts 1b2, 2122, 1b2* and 2122*. The two contacts 1b2 and 2b2-constitu-te a double contact that closes for posi tion I, that is, for a particular direction of the current in winding B2, while contacts 1b2* and 2b2* close when the direction of the current is reversed, that is, for position II. As is seen, contacts *1b2 and 1b2*, and M2 and 2b2*, respectively, are interconnected in such a way as to set up the same connection whatever the direction of the current is in B2. 1

The second relay has two windings 1B1 and 2131 wound in opposite directions and ordering two' double contacts 112 1, 2b1 or '1b1*,'2b1*, respectively. The third 'relay has three windings 1B0, 2B0, 3B0 to control two double contacts 1b0, 2b0, or 1b0*, 2b0*, respectively. The fourth relay is like the third, but. the index 0 is replaced by u.

The digits of a denomination row are defined by the closure of these contacts according to the table below:

contacts contacts I Digits It will be seen that the ten digits of a denomination row can be defined by a set of four relays, of which two always function simultaneously. Thus, the number of relaysprovided for each denomination row of .the B group is equal to half the number of positions necessary to define all the digits of a. denomination row, rounded out to the next higher whole number. The relation, as indicated above,

neonates between. the making of the contactsand. the. number that is defined makes it possible. to convert. simply tothe nines. complement. For, the said relays are polarized and. show three positions, as. has already been. explained, to wit. a. rest position. 0, a position I'. for which. contacts bi are closed and aposition II for which. contacts. bi are closed. If contacts bi or bi*', respectively, ofa relay are closed for a given direction of the control and polarization currents, the direction of one of these currents need only be reversed for contacts Iii to be closed instead of'lii, or contacts bi instead-of bi*, as-the case-n-ray-be; Referring to thetable above, it. will be seenthat reversingthe. direction of a current in" therelays of the B group' suffices to cause the closure ofthe contacts defining, not. the number transmitted, but its nines complement; It will be" seen, for example, that the number. 3 is. defined by the closure of contacts lbu, Z-bu and"1b-1* 2111*; If the direction of the control current, for example, is" reversed; that will bring: about the closure ofzcontactsi sl'lzu zbui and: 15b1, 2121,, which define: the: number: 6. We; have thus: carnied out, for this. denomination row, the snbtractiom 9'-'-3=6. The rest: of the: description: will show'thal: this possibility has many advantages and is made use of to; make calculatin g: operations possible.

Two of the four' relays have a. third: controli Winding 3'39 or 3B'u, respectively These'windingszare provided to make it possible to introduce the number zero: by means of. an. independent control. circuit. B'y reversingthe direction. the. current. in these. suppiementary windings; those contacts may also: be-rna'de close that define the number nine instead: of. zero.

Referring again to FIG- 11);. the calculation circuit has two inputconductors ge and gel, which are: connected to five double contacts. 1'b0;. 2&0, lbl, 2B1, 1b 2; 2b2, 151*, 2121*, lbtli 2110*. These: contacts. correspond to those represented in FIG. 12; butcontact's 1 52'? and 2152*, which are connected in parallel to contacts 1b2 and Zbl, have not. been shown in. order to: avoid overcrowding. the draw.-

ing. These contacts ofthe B: group are connected. to six movable points: 1am, 2am 6am of a.multiple-contact wiper. These six movable. points can he shittedlwith respect to ten fixedipointsrlafllafi 10af,.which are connected to=windings 1R0,- 1111,1322; 1R3-,. -1R4,,2R0,L 2R1, 2R2, 2R3, 2R4, respectively, ofifive relaysof the R group. These windings are mounted two. on a: relay. Thus, 1R0 and 2R0 belong to a: single relay; likewise: for IE1: and 2R1, 1R2: and 2R2, etc. The above-mentioned contacts of. the B group candefine. adigii: from.0ito 4.. Themovable points of the multiple-contact wiper can also-assume five different positions with respectto the hired points, so as to: define thed-igits from. 0'to.4.

The. totality of thesecontacts' of the A and B groups makes up an addition circuit, sothat. a. current entering at ge goes only through thatcontrol winding of relay R thatxcorrespondsito the result of the: addition of the numhers. defined. by the. contacts ofthe A and 3 groups. In the first denominationrowdoublecontact:1171*, 2191* is closed and. defines the digit..3,.while. the movable points of the multiple-contact wiper are shifted two positions up and define the digit 2. If a current is sent throughinpu-t. con.- ductor ge, it will. go through winding 2R0 after passing through contact 1711* and the multiple-contact wiper. Winding 2R0 enablesusto define the digit 0 or the digit '5 in the R group, according. to the positions of relays. Ru and Rv, which are connectedzin series inthe same denomination row. That is, windings-1R0 and 2R4 form two groups ending in two output conductors. Windings 1R0 to 1R4 are connected to one of these output conductors and windings 2R0 to 2R4 to the other. These two output conductors are connected to" two double contacts lbu, 2bu orlbu 2bu*', respectively, that have been'described in connection with FIG. 12". These contacts are connected to a multiple-contact wiper having three movable points Iamu, Zamu, 3amu, capable of assuming two positions with. respect to four fixed points 1afu, 2afu,,3.afu, 4afu.

8 These fixed points are; connected to. four. windings 1Ru, 2Ru, 1Rv and 2R v of two relays Ru and Rv. This calculation circuit. offers two output conductors for this denomination row. Windings 1Ru and 1R1: are connected to one of these output leads and. windings 2Ru and. 2R-v to the other. I

A current passes through winding l-Ru when the: result or, thedigits in. thi denominations-ow defined in. groups A.and.,B is lessthautfive. Whemthisresultisbetwcen. 5 and. 9, the' current passes through lRv. When the result ofthis addition isat. least equal to ten, the current leaves the first denomination row by the other outputconductor, passing through ZRu it theresult. is between 10. and. 14 and throughZRv it the. resultis between 1.5. and 19.

When the current. enters. through-gel, the result. indicated by relays R is increased by a unity compared to the result of the addition of the numbers defined by the contacts of the A and-B groups.

Decimal places 2' and 3 are identical withthe first denomination row and the two input conductors of a denomination row are connected to thetwo output conductors of the" preceding denomination row; Decimal places 4', 5' and 6 are: analogous to decimal places I, 2 and 3-, but only" have contacts oft-he B group, do multiplecontact wiper (-group A)-' being provided for these denomination rows, since the keyboard oi? themachine has only thre'e'columns. Therelays R ofdecim'al places- 4, 5 and 6 haveonly one main winding, however, with t-he encep-- tion of relays R0 and Ru; each or" which has two wind ings, 1-R0 and 2R0, or lRu, 2Ru, respectively. 'Thissi'm plification isdue tothe fact that the largest figure that can occur in a denomination rowis equal to 10" (that is; 9*plus a carry), since no A'|B additions occur inthese decimalplaces. The windings that only serveto define numbers greater than 10- would therefore be' superfluous and-have been eliminated. All the decimal places-are connected in series,; one after the other, so that the current entering through go or gel goes through all of them before leaving by go. If the sum of the two numbers dcfinedby'groups A and B in a denomination row is smaller or larger'than ten, the current leaves by oneiorthe' other of the output conductors of this decimal place, producing or not producing a supplementary addition of a 1' in the following denomination row, accordingly.

As the current goes through the windings of certain relays of the R group, as indicated above, it causesthev closure of the contacts of those relays that are. part of the transfer circuit represented in FIG. 11'. Each relay of theR group has'two. contacts riand ri* that allow the current to be sent through the windings of certain relays of the B group. Dependent on the direction of-the current during'the calculation phase, thatis, of the current going through the calculation circuit of FIG. 10, the windings of relays R will be traversed in one direction or the other and cause the closure of contacts ri or contacts ri Referring to FIG. 11, it is seen that the contacts r1: of a denomination row are connected to the control windings of the relays B of' the same denomination row; on the other hand, contacts ri* of a denomination row are connected to the control windings of'the relays B of the'foll'owing decimal place.

Contacts ri. or ri*, respectively, of a denomination row are connected to relays B of the same denomination row or the following. denomination row, respectively, so as to send the current, during a transfer phase, through the windings of those relays B that define the same number as the R relays.

By reversing the direction of the current during. the trans-fer'phase, that is, by having it enter at fa and leave by fe, those contacts ofthe-B group are made to close that d'efi'ne the nines complement. of. the number defined by the contacts of'the R group. Finally, the. transfer circuit has an independent. second. circuit with an inputconductor file and. an output. conductor f0aand comprising all thewindings 3B0. and;5Bu.. If, instead ofhaving'thecun