Apnl 10, 1962 WEN TSlNG CHOW ETAL 3,028,659

STORAGE MATRIX Filed Dec. 27, 1957 MI 85 g; IO' al l 8| NIH-f so In 82 yvvavraxw.

L WEN TSING CHOW 2 ml 12m WILLIAM H.HEN12l-C.H K 20 l' ,2 BY f ll2a l22a I 47 H2 22 F" -loa, 2o 1\ 2041! 45" A 7' TOP/V: 5

Unite States Woodbury, N.Y., assignors to American Bosch Arma Corporation, a corporation of New Y orlr Filed Dec. 27, 1957, Ser. No. 705,555 6 Claims. (Cl. 29-4555) The present invention relates to digital computers and has particular reference to matrices used therein.

In digital computers, the numerical constants appearing in various equations are set in by devices known as matrices which essentially comprise a gridworl; of crossed electrical conductors, each electrically insulated from the others, with an electrical element such as a semiconducting diode or rectifier, for example, connected b tween conductors at selected cross-over points. Pulses are applied in timed sequence to the input conductors, and a digital number in serial form is produced at each of the output conductors. The digital number so produced is normally employed as one of the constant values required in the solution of a mathematical problem by the digital computer.

Matrix construction may be a problem in field use or where the computer must accept the various constants at a given instant when these constants are continually changing. Under such conditions, the equipment and time to prepare the required matrix in the normal fashion may not be readily available and, furthermore, the opportunity for rigid inspection may be severely limited.

In order to provide for field reliability, flexibility and possible remote preparation of matrices, the present invention proposes to prepare a matrix having an electrical element at each cross-over point in the laboratory or shop and then to remove unwanted elements in the field by physically disconnecting or electrically burning out the unwanted elements. The improved reliability stems from the fact that each connection can be made and inspected undershop conditions. The flexibility is afforded by the speed at which new constant boards can be prepared by blowing out certain connections, and is of particular value where certain data becomes available just prior to the start of calculations. The remote setting is made possible by providing connections from a control station to the'matrix boards in order to burn out the undesired elements at the remote location. These connections may be by electrical cables or, conceivably, by wireless means.

For a better understanding of the present invention, reference may be had to the accompanying diagrams, in which FIG. 1 illustrates a matrix board of preferred construction,

FIG. 2 illustrates a matrix board in use, and

FIG. 3 shows a double matrix which is used for a particular type of non-linear device.

With reference now to FIG. 1 of the drawings, a series atent of parallel conductors 1, 2, 3, 4, 5 are stretched between the ends of an open frame 6, and a second series of parallel conductors 10, 2t), 30, 40, 50, 6t), 70 are stretched between the top and bottom of the frame 6. The first conductors 1 through 5 are on one face of the frame 6, while the second set or" conductors 1G, 20, 3t), 40, 5t), 60, 70 are on the opposite face of the frame 6 and are substantially perpendicularly disposed to the first set of conductors, where each conductor is insulated from all the rest of the conductors.

' At each of the cross-over points, an electrical element such as a diode for example, is connected between the conductors. Thus, diode 11 is connected between conductors 10 and 1, diode 12 is connected between conductors 10 and '2, diode 13 between conductors 10 and 3, etc.

A matrix board may be used as shown in FIG. 2, for example. A series of timed pulses are applied sequentially to the conductors 1, 2, 3, 4, 5, as represented by the pulses shown to the left of FIG. 2. Assume that constants having values of three, six, nine, :twclve, fifteen, eighteen and twenty-one are required in binary notation for certain calculations at conductors 10, 20, 30, 40, 50, 60, 70 respectively. The matrix is accordingly adapted to allow the first and second pulses to be transmitted to the conductor 10 through diodes 11 and 12 so that a binary representation of three appears at the conductor 10. Similarly, the second and third pulses are transmitted to conductor 20 through diodes 22 and 23 to produce a binary representation of six at conductor 20. The first and fourth pulses are transmitted to conductor 30 through diodes 31 and 34, third and fourth pulses are transmitted to conductor 40 through diodes 4'3 and 44, the first four pulses are transmitted to conductor 50- through diodes 51, 52, 53, 54 and so on.

In prior methods of constructing the matrix, only the diodes 11, 12, 22, 23, 31, 34, 43, 44, 51, 52, 53, 54, 62, 65, 71, 73 and 75 would be soldered between the respective conductors toprovide the desired transmission characteristics. However, as pointed out earlier, the soldering may have to be undertaken under unfavorable conditions, resulting in relatively unreliable connections and requiring considerable time. The method of the present invention, however, requires that the extraneous diodes of the matrix in FIG. 1 be removed, to provide the desired transmission characteristics. Thus, diodes 13, 14, 15, 21, 24, 25, 32, 33, 35, 41, 42, 45, 55, 61, 63, 64, 72 and 74 must be removed from the prepared matrix of FIG. 1.

They may be removed simply by clipping the leads across the undesired diodes, for example. With miniature components, however, where the space is limited, the physical clipping of diode leads may cause injury to other elements. Also, where the matrix is encased in plastic for protection from environmental conditions, there may be no physical access to the diode leads.

Therefore, the preferred method of removing the diodes is by burning them out electrically. The term burning is intended to denote that the electrical element is.

subj ected to a high current such that the electrical element fails, and an open circuit is created across the leads of the electrical element. A diode for example, maybe burned out by connecting a high reverse voltage across its leads to thereby cause disintegration of the diode junction and to produce the desired open circuit. Under certain conditions destruction of the diodes may be undesirable, and for this reason a fuse, which will burn out before the diode will, may be connected in series with each diode. The term blowing out or burning out further includes any process which, by means less drastic than actual destruction of the non-linear elements, effects a change of the circuit impedance to a level which makes the particular circuit inoperative. Such means may create a change in the magnetic characteristics of the circuit element to cause inoperativeness, for example.

For ease of production, the diode matrix may be con nected to the rotary switches and 81 as in FIG. 1. The conductors 1, 2, 3, 4, 5 are connected to the stationary contacts 1', 2, 3, 4, 5, respectively of switch 80 through I the wires 8-2. A plug 83 and sockets 84 are provided to permit easy connection and removal of the matrix from the switch 80. Similarly, the conductors 10', 20, 30, 40, 50, 60, 70 are connected to the stationary contacts 10, 20', 30", 40', 56', 60" and 70' of the switch 81 through wires 85, while plug 86 and socket 87 are provided to permit easy connection and removal of the matrix from switch 81.

1 o In the position of the switches 80 and 31 shown, it will be seen that the diode 63 is connected across the movable then be removed from the circuit of FIG. 1 and inserted into a computing circuit elsewhere, as required.

It will be apparent that many alternative arrangements maybe provided for particular conditions. For example, if the physical location of the matrix board precludes convenient access thereto, selector switches may be provided in place of the plugs 83 and 86. In one position of the selector switch, the matrix would be connected to switches 36 and 81, and in the other position of the selector switch, the matrix would be connected to the computer. Further, the switches 80, 81 could be stepping switches, for example, controlled from a remote point for selecting the undesired diode, and switch 91 could be a remotely operated relay for applying the power supply 99 to the diode. Such remotey operated switches could be connected to the remote control station by electrical cables, or by wireless means where distances are too great for the direct connection.

In addition, it should be realized that although the description has specified that diodes are connected across the conductors, other non-linear elements can be used equally well. The other non-linear elements may inciude such devices as magnetically biased ferrite cores for exampie. In these devices the state of permanent magnetism of the magnetic core'di ctates the impedanccs of the device to currents of opposing polarities. in a nonmagnetized state the device is linear. When magnetized in one direction the impedance to pulses of one polarity will be greator than the impedance to pulses of the opposite polarity. The magnetization of each core is controlled by a coil so that the magnetization control circuitry will require a separate matrix from the one to which the ferrite cores are connected. The principle is illustrated in FIG. 3 where the ferrite cores 111, 112, 121, 122 are connected between the wires 1 and it 2 and 1d, 1 and 20, and 2 and respectively to form one matrix. The control circuits, shown as coils 111a, 112a, 121a and 122a, are connected to a separate matrix including the wires in, 2a, 10a and 20a in a corresponding fashion. The wires 1a, 2a may be connected through leads 82 to the switch 80 as in FIG. 1, and wires 10a, Zila may be connected through leads $5 to the switch 81. Accordingly, the control coils of those ferrite cores which are to be made inoperative are selected by the switches 84 and S1 and upon closure of key 91 the selected ferrite core will be magnetized to the desired non-linear condition.

The use of the double matrix, one connected to the non-linear impedance elements, the other connected to control circuits for the non-linear elements, is not limited to the magnetic devices described. It may be used wherever the impedance characteristics of the non-linear ele-- ments are controllable .by separate electrical means. Other non-linear devices will occur readily to those skilled in the art without further examples being listed here. The five by seven array shown in FlGS. 1v and 2 has no particular significance and was chosen merelyior illustrative purposes. The matrix used in an actual computer usually would be many times larger than the matrix shown in the figures.

conductors, and later removing such of said electrical elements as are not required.

2. In the method for producing a matrix board for digital computers, the process of first'makinga matrix board including a first-series of parallel conductors, posi tioning a second series of parallel conductors perpendicular to the first series of conductors, connecting a nonlinear electrical elernent between each of said first series of parallel conductors and each of said second series of said conductors adjacent each cross-over point, and later removing such of said electrical elements as are not re quired.

3. In the method for producing a matrix. board for digital computers, the process of first making a matrix board including a first series of parallel conductors, positioning a second series of parallel conductors laterally of the first series of conductors, connecting a non-linear electrical element between each of said first series of parallel conductors and each of said second series of said conductors, and later removing such of said electrical elements as are not required by physically disconnecting said elements not required.

4. In the method for producing a matrix board for digital computers, the process of first making a matrix board including a first series of parallel conductors, positioning asecond series of parallel conductors perpendicular to the first series of conductors, connecting a nonlinear electrical element between each of said first series of parallel conductors and each of said second series of said conductors adjacent each cross-over point, and later removing such of said electrical elements as are not required by physically disconnecting said elements not required.

5. In the method for producing a matrix board for digital computers, the process of first making a matrix board including a first series of parallel conductors, positioning a second series of parallel conductors laterally of the first series of conductors, connecting a non-linear electrical element between each of said first series of parallel conductors and each one of said second series of said conductors, and later removing such of said electrical elements as are not required by electrically burning out said elements not required.

6. in the method for producing a matrix board for J digital computers, the process of first making a matrix board inciuding a first series of parallel conductors, positioning a second series of parallel conductors perpendicular to the first series of conductors, connecting a nonlinear electrical element between each of said first series of parallel conductors and each of said second series of said conductors adjacent each cross-over point, and later removing such of said electrical elements as are not required by electricaliy burning out said elements not re quired. t

References lite-d in the file of this patent UNITED STATES PATENTS 2,399,753 I McLorn May 7, i946 2,694,249 Kapp Nov. 14, 1954 2,769,865 Faulkner Nov. 6, 1956. 2,769,968 Schultheis Nov. 6, 1956 2,918,669 Klein Dec. 22, 1959