US2435840A - Computing device - Google Patents

Description

Feb. l0, 1948. s. A. MoRToN 2,435,840

COMPUTING DEVICE Filed nec. 28, 1943 5 sheets-Sheet 1 if@ Z.

llllllll lillllllllllllllllllllllll Illlllllillllllllllll lllllllllllllillllllllll llilllllllillllllll IIIIIIHWJHIIIIIIIIIII llllllllllllllllllll IIIIHIIHIH Snoento EEUREE- FLM nR'mN Gttdnug Feb. l0, 1948. G. A. MoRroN COMPUTING DEVICE 5 Sheets-Sheet 2 Filed Dec. 28, 1943 r N .m m RL N am Hw H D/ R i n m E lf 1 00 00000 u u v m #im 2-... 1 Y f n a n m 4 m 1.1.1.1 l n 0 u n T0000 t. n n n m z n 1: u?1 11 nudi u l u..." .I Y 6 0 m mz j l F IIL u n l. cd MN M w F, lllll 1. m f

attorney;

Feb. 10, 1948. G. A. MoRToN COMPUTING.v DEVICE 5 Sheets-Sheet 3 Filed Dec. 28, 1943 Feb. 10, 1948. G. A. MoRToN `COHPUTIING DEVICE 5 Sheets-Sheet 4 Filed Dec. 28, 1943 EURBE H. MURTDN G. A. MORTON COMPUTING DEVICE Filed Dec. 28, 1943 5 Sheets-Sheet 5 Snoentor EEURBE HMDRTIJN attarneg *fPatentes:Feaffiof COMPUTING DEVICE George A. Morton, Princeton, N. J., signor to Radio Corporation of America, a corporation of Delaware Application December 28, 194.3, Serial No. 515.887

6 Claims. (Cl. 23S-61) functions of two variables such as are required to determine time of flight, gun elevation and the like consists basically of determining the value of the function and its rate of change with respect to the two variables at a number of discrete points and interpolating between these points. In accordance with the present invention, matrices are used for evaluatingthe required quantities, and the fundamental elements containing the matrices for generating impulse number trains representing the necessary quantities are cathode-ray tubes, similar to the monoscope utilized in television technique.

The monoscope consists of an electron gun, deection means and a special screen or target which is connected to an external signal lead. These elements are enclosedin an evacuated y envelope. Infomation is recorded on the special screen in the form o! line lines of carbon or the like, having a different secondary electron emission rate from the screen background which may consist of aluminum or an'equivalent material. As the electron beam sweeps across the screen, a

' current pulse due to the change in secondary electron emission is generated each time a line is crossed. Pulses formed in this way are amplied and applied to a totalizing device.

The screen may be similar to a sheet of cross section paper in that it is divided into squares or areas so that the horizontalrows represent one set of factors of terms and at least one term of the fimction. Information is recorded in the squares in the form of lines representing binary numbers. Two types of linesiare used, a narrow line for digit zero and a wide line for digit one involvned mathematical v (i) The width oi the electron beam is made just V small enough to resolve the wide lines.' Consequently, the output pulses are ofv the same width but differ in amplitude, a large amplitude representing 1 and a small amplitude representing 0.

' When the electron beam sweeps across such a group of lines, potentials representative of a series of digits are developed.

Thus, to obtain one set of values of the function, the beam is deflected vertically to the horizontal, row of successive areas on which are recorded values of the function for the selected valueof one variable (i. e., y). For this step, it is arranged that the vertical deflection is linear andthat Vo-YVi volts are applied to the vertical-deflecting electrodes, Vo being the voltage required to deflect the beam from the center to the bottoml of the matrix pattern and Vi the voltage required to `move the beam one row vertically. 'I'he beam is then deflected horizontally to scan the row of areas on which are recorded the selected set of values of the function.

The signals from the selected row of screen areas are in the form of potentials representative of the successive digits of the binary numbers recorded on that particular row. These'potentials are amplified and then divided into stepping pulses (one for each digital position) and impulses corresponding to the digital positions where 1 appears in the recorded numbers. The stepping pulses are utilized to control the operation of a totalizer wherein the desired value of they function is established, the pulses corresponding to the digit 1 being utilized as a multiplier or merely added to the sum then in the totalizer as required to establish the function. As hereinafter explained, a desired number of zeros may be recorded preceding or following certain of the matrix numbers for facilitating these operations of the totalizer.

Important objects of the invention are to provide means for generating a function having certain factors of some of its terms and one of its terms recorded in a single row on a matrix, to provide means for deriving different values of the function from diierent rows of the matrix and to provide means for selectively establishing each value of the function in a tot-alizer.

The invention will be better understood from of .values of these quantities.

the accompanying drawings considered in connection with the following description, and'lts scope is indicated by the accompanying claims.

Referring to the drawings:

Fig. l is a block diagram of the entire computing system,

Figs. 2 and 2a show details of the monoscope screen or target on which the different values of each set are recorded in different horizontal rows,

Fig. 3 is a wiring diagram of the control circuits cf the computer.

Fig. 4 is a wiring diagram of the circuit by which signals derived from the monoscope screen are separated into stepping pulses and pulses corresponding to each digital position on which a digit l is recorded,

Fig, 5 is a wiring diagram oi' the two right hand I units of the totalizer, and

Fig. 6 illustrates four units of the set-up devices utilized to select a predetermined row oi monoscope screen values and to establish the value by which other values are multiplied in the totalizer.

In considering the various figures. it should be understood that all the signals derived from the monoscope screen are supplied through a single lead to the amplifying and separating circuit of Fig. 4, that the stepping pulses derived from this circuit are applied to -each A unit of the totalizer to advance the number in the totalizer one digital position to the left, that the pulses representing the digit 1 are fed into the lrst of the totalizer to be added to the number. if any, already in the totalizer, and that, at certain stages of the operation, the number in the totalizer is fed into the transfer units T for multiplying it by a' number previously established in the X set-up unit.

If the function to be generated is the number my) and the coefficients a(y) bw) and ely) are recorded along the yth line of the screen so thatl they are read from left to right as c, b, a and fo. The value of the variable y is used to select the y (th) row of the screen areas and the value :i: is used as a multiplier in the totalizer. The same procedure is followed with respect to al1 the other horizontal rows of values recorded on the screen.

The computer of Fig. 1 includes a cathode ray device I which is provided with an electron gun I3 for forming a beam of electrons, a pair of electrodes Ill for deiiecting the beam in a vertical direction, a pair of electrodes I for deilecting the beam in a horizontal direction, and a screen or target I6 lupon the horizontal rows of areas of which are recorded successively values of c, b, a andA f, there being one row of areas for each set Secondary electrous emitted by the target I 6 are collected by the coating I6 which is maintained slightly positive with respect to the target IG as is customary in the operation of cathode ray devices.

The vertical deflecting system of the cathode ray device I0 includes a switch SV having three leads. Through one of these leads, .pulses are supplied to the switch from an impulse generator.

Through another of these leads. these pulses are applied to the y set-up unit which may be of a well known trigger circuit type for adding to the number set up in it and to an amplifier-rectifiercapacitor unit 60--6I--62 for deilecting the beam to a horizontal screen row of areas predetermined by the number set up in the y set-.up unit. When the number in the unit reaches zero, a resulting positive pulse is transmitted from one lead of a trigger circuit unit of the set-up device to the switch through its other lead for interrupting the transmission of pulses and stopping the vertical deflection at the selected row of areas.

A. negative pulse produced at another lead of such trigger circuit unit by clearing of the coarse y set-up device is also'utilized to initiate operation of the horizontal deflection system II and, through a switch l2, the positive pulse similarly generated is utilized to activate the amplifier and separation channel Il.- The stepping pulses delivered from this. channel are supplied through a lead I9 to each A unit of the totalizer AB. The pulses representing the digitl 1 are supplied to the right hand A unit of the totalizer through a lead 20.

As hereinafter explained in connection with Fig. 5, the number establshed in the :c set-up unit is transmitted to the totalizer as a result of a stepping pulse on lead I9 only when a potential representative of the digit 1 is transmitted from the totalizer through the lead 2I and an amplifier 22 to the transfer units T. The totalizer units A are trigger circuit units of a well known type which includes two cross-connected electron discharge tubes. only one of which conducts current at any given time. Transfer of current from one to the other of these tubes is effected by the appli cation of a negative pulse to the common anode resistor terminal or symmetrical point of the two tubes or to the control grid of the particular tube that happens to be conducting when the negative pulse is applied.

The totalizer units B are similar to the A units with the exception that the grid resistances of the two tubes of each unit are made unequal so that the application of a negative pulse to the grid of the tube having the highest grid resistance causes current to be transferred to the other tube of the unit only for a very short time, after which it is automatically returned to the tube having the higher grid resistance. These B units are connected between the A units for carrying over a digit 1 from a lower to the next higher digital position. Thus the unit B1 carries over ones from the unit Ai to the unit Aa, the unit B: carries over ones from the unit A: to the unit A3, etc. The delay circuit in the lead 20 is for the purpose of so timing the digit pulses on lead 20 from the amplifier separator that they do not interfere with the pulses applied to the A units through the transfer units T.

Assuming that (1) the y and :c values corresponding to a particular horizontal line of screen areas have been established in the y and :c set-up units, and 2) the electron beam has been vertically delected to this line, all as previously indicated, the operation oi' the system is apparent from the following tabulation. In this tabulation, the first column shows the successive operational steps by Which the selected value of the function is transferred to the totalizer. 'I'he fourth column shows the transfers of the digit 1 through the lead 20. The third column shows the transfers of :c through the transfer elements T. The second column shows the stepping pulses through the lead i9. The ilfth column shows the number in the totalizer A units which represent successive digital positions of the number, lowest l digital position being at the right (A).

For the purpose of this sample calculation, it

is assumed that the diilerent values are expressed in the binary numerical system and that x=l0, f11=1011, a=101, b=110, and c=11. A pulse on any lead is indicated by 1 and the absence of a l 6 6. :rfc-l-:rb-i-a is multiplied by x and at the same time f is added to the total, giving the desired number, fiy) =:1:=c+:1:2b+za+f.

It will be noted that the essential computing It will be noted that the operation involves the following steps:

1. :c is set up in multiplicand of multiplier unit.

2. Beam is deflected vertically by the y value to the desired horizontal row containing c, b, a and fn.

3. cis fed to totalizer unit through the lead 20.

4. c is multiplied by :z: by being fed to the transfer tubes (from B11 and ampliiier 22) as a result of being stepped along by stepping pulses of b. (Note the actual number b is preceded by enough stepping pulses to adjust totalizer to bring c to the proper position.) At the same time, b

1s added to the product. the digit 1 pulses being delayed so that each digit 1 arrives just after pulse on any lead is indicated by 0. 5 sequence at the totalizer is always: stepping pulse 1.0111110 ad operating Lead 21 1111111 1 No.1n

Steps sgfg Transfer from Totalizer mmh tube a 7 0110000000 111111111 nepping pulses to adjust totailzer 5 1 1000000010 step di laos 1 1000000101 mdded 0 0000001100 7 1100000000 initial stepping pulses 11 1 0000011100 w+bz+11 12 5 1110100000 initial stepping pulses 17 i 0001000010 z=+bz1+u+f1 on lead I9 followed by transfer pulse on lead 2|, followed by digit l pulse on lead 20. In any particular case, the transferred digit may be 0 and there will be nopulse on lead 2l or the digit on lead 20 may be 0 and there will be no pulse on lead 20, but the timing is always the same.

Figs. 2 and 2a show front elevational and sectional views of the screen I6. The details of the screen have been explained above.

The details of the manner in which these various steps are produced will be readily understood from a consideration of the wiring diagram of Fig. 3, wherein the various parts are enclosed by dotted lines and are generally indicated by reference characters applied to these dotted enclosures. In this gure, DV is the vertical deileotion generator. G is an impulse generator from which pulses are applied to the vertical deflection generator DV for moving the electron beam vertically and to a y set-up unit S for reducing to zero a u number previously established in the set-up unit. SV is a switch Ywhich ter- 7 minates the supply of impulses from the generator G when the number established in the unit S is reduced to zero. DH is the horizontal deflection generator for moving the electron beam to scan the selected horizontal areas. I2 is the switch through which the amplifying and separating channel between the monoscope screen or target I6 and the totalizer A--B is controlled.

The complete vertical deflection system includes the switch SV (comprising the trigger circuit tubes 54 and 65) and a control tube 56. Positive pulses are applied from the generator G through a capacitor 58 to the control grid of the tube 56 which is capable of delivering negative pulses to the y set-up device and to the deection generator DV only when the'tube 54 is conducting and a positive potential drop exists in the cathode resistor 59 of the tube 54.

Assuming the tube 55 to be conducting, the negative pulses supplied through the tube 55 function to subtract from the count-set up in the y set-up device and at the same time (through the amplifier 60 and rectifier 6|) to increase the charge of a capacitor 62. As the charge of this capacitor increases, the voltage between the vertical defiectors Il is increased and the beam is moved to the line of areas containing the 4pertinent information. Then the y counter clears and thereby transmits a positive impulse to the control grld of the tube 55, thus transferring current from the tube il to the tube 55 and biasing ofi the control tube 56. The vertical deflection system is restored to its initial condition by the application of a positive pulse to the lead 63.

The horizontal deilection generator DH is an ordinary saw tooth generator sometimes used in connection with television circuits.` It includes a capacitor 23, which is charged through a lead 2t and is discharged through a tube 25 in response to a negative pulse, supplied upon clearance of the y set-up unit, through a lead Sli to the control grid of a direct current ampliiler 26. The cathode-anode potential of the tube 26 is normally such that it conducts sufcient current to maintain the tube 25 at cutoi. The application of a negative pulse to the grid of the tube 25 cuts off its current. This makes the grid of the tube 25 more positive, thereby discharging the capacitor 23 and producing a very rapid initial deflection of the beam to the start position. After the negative pulse applied to the grid of the tube 26 subsides, the tube 25 assumes its normal cut-off condition. Thereafter charging of the capacitor 23 results in a gradual rise of the potential at the upper plate of the capacitor 23. The resulting change in potential is applied through a coupling capacitor 2l to the single-ended push-pull amplier stage 28-29, and thence to the horizontal deecting electrodes I5 of the cathode ray device I0, thus deiiecting the beams along the selected row of screen areas on which the pertinent data is recorded. g

At the same time that a negative pulse is applied to the horizontal deflection generatori DH, a positive pulse is also applied through the lead 93 Ato the control grid of the left hand tube 3i of the switch I2, thereby producing in the cathode resistor 52 of this tube a positive potential drop which neutralizes the negative potential of a battery 53 connected in a control lead of the amplifying and separating channel and permitting signals to pass through this circuit to the totalizer A-B. How the negative and positive control pulses are produced by the y set-up device S is previous explanation.

When the y counter has cleared, the area containing the pertinent data is scanned and the resulting signals are applied through the lead 8l to the control electrode of an amplier 8B and are further amplified vby the ampliers B'I and Il. From the output of the ampliiler 68, the signals are supplied to two paths, one of which includes a limiter unit 69 and a shaping unit 'I0 for producing the stepping pulses and the other of which includes an amplifier unit 1I, a clipper unit I2 and a shaping unit it for providing the counting pulses which represent the digits l of the recorded data. The character of the signal pulses at each stage of their progress through the circuits is shown above the circuits and is readily understood in view of the fact (previously explained) that each pulse represents a digital position in a binary number, the high amplitude pulses represent the digit 1 and the low amplitude pulses represent the digit 0. The stepping pulses appearing at the output terminal I9 are applied to the A units of the totalizer and the counting pulses appearing at the terminal 20 are delivered to the right hand A unit of the totalizer, as previously explained.

Fig. 5 illustrates the wiring of two units of the transfer and two units of the totalizer A-B.

At this point, it should be understood (l) that the trigger circuit units of the switches SV and l2 (Fig. 3) are of the type in which only one o! the cross-connected tubes is conducting at. a time and current is switched from one tube to the other by the application of a negative pulse either to the grid of the conducting tube or to the upper terminals of 'the anode resistors of the two tubes, (2) Athat the y and z set-up units include similar trigger circuit units which'are connected in -cascade as indicated in Fig. 6, (3) that the units A of the totalizer A-B include a number of trigger circuit units, and (4) that the units B of the totalizer are similar to those of the units A with the exception that they are of the slideback type, i. e., the grid resistance of the right hand tube ls made higher than that of the left hand tube so that a negative pulse applied to the grid of the right hand tube transfers current to the left hand tube after a slight delay and only for a very short time, after which current is automatically returned to the right hand tube. These various set-up circuits are more fully disclosed and claimed in my copending application Serial No. 473,146, -led January 21, 1943, now Patent No. 2,404,047.

The circuit oi Fig. 5 includes a number oi.' transfer tubes 15,11, etc., there being one such tube for each digital position of the number m. It also includes a trigger circuit unit A1, Az, etc., for each digital position of the number representing the generated function, Connected between the A units are the slide-back units Bi, Bz, etc., which transfer a digit 1 (from one A unit representing one digital position to another A unit representing the next higher digital position) when the lower digital A unit changes from a asesinato binary 1 to a binary 0 condition. Potentials representative of the quantities .1: areapplied through the leads .18, 19, etc., stepping pulses (one for each digital position) are applied through the lead I9 and pulses corresponding to the digits 1 in the number recorded on the monoscope screen are applied to the lead 20. Thus,-

when the right hand tube is conducting, the i unit is in a binary zero condition and its indicator 80 is unlighted and, when the left hand tube 'is conducting, the unit is in a binary one condition and its indicator 80 is lighted.

Assuming the totalizer units A1 and A: represent successive digital positions and that unit A1 is in a binary one condition with its left hand tube conducting, the application of a negative pulse to the A1 unit will convert the unit to a binary zero condition, thereby applying a negative pulse to the grid ofK the right hand tube of the slide-back unit B1 and temporarily transferring current from its right hand tube to its left hand tube. This tube remains conducting a predetermined number of microseconds, then the conduction reverts back to the right hand tube. When this occurs, a negative pulse is transferred through the couplingcapacitor 82 to the anode resistors of the unit A2. If this unit is in a binary zero condition, all that happens is the transfer of current to its left hand tube. thereby converting it to a binary one condition. If it is in a binary one condition, it is converted to a binary zero condition and ar negative pulse is applied through a capacitor 83 to the control grid of the right hand tube of the slide-back unit B2, thereby temporarily transferring current to its left hand tube and upon its return applving through a capacitor 84 a negative pulse whereby a digit 1 is transferred to the Ae unit (see Fig, l). This, process continues throughout the successively higherdigital position units of the totalizer, as has been explained above. Between these transfers of potential from the leads 18, '19. etc., a negative pulse is applied through the lead I9 and the capacitors 85, 86, etc., to the grid of the left hand tube of each A unit for stepping the accumulated number to the next highest digital position;

The effect of a negative pulse applied to the lead I9 is readily understood if there be assumed to have been established in the units A1. Az, A3, etc.. a representation of the binary number 11010. Under these conditions, the negative stepping pulse has no direct effect on the Ai and A3 units, which are in a binary zero condition, for the reason that the negative pulse is applied to the grids oi?` tubes which are not conducting. The A2, A4 and A5 units, however, are in a binary one condition and the negative stepping pulse functions to convert each of these units to a binary zero condition. When this occurs, the unit Bz functions to convert A3 to a binary one condition, the unit B4 functions to convert A5 to a binary one condition and the unit B5 functions to convert Ae to a binary one condition. As a result, there is now established a representation of the number 110100, i. e., the number 11010 has been stepped one digital position to the left.

Fig. 6 is a wiring diagram of the y and :r set-up units. As already indicated, each of these set-up units includes a plurality of trigger circuits 8'5-86, IVI-88, 89?-90, 9I-92, etc. (one of each digital position in the highest value of the particular value to be established). If each of these circuits is in a binary zero condition (right hand tube conducting), the value required to clear the counter at the desiredl valuefoi' w forexample, is` set up by applying a negativepulse to the termi-A nals 99 of such circuits as are reoulrerlitoV be In oi a number of'pulses determined by the value of y. Thus it-the eighth line ofthe target I 8 is to be selected, a negative-pulse is applied to the lead 95 of the trigger-,circuithnit 9I-192, With -current conduction in the left hand side of this unit and in the right hand side 0i the units -86, 91-88 and 89-90. the application of eight negative pulses to the lead 9B causes current conduction to be in the right hand sides of all the trigger circuit units. thereby producing a positive pulse at the lead 93 and a negative pulse at the lead 94. Likewise, if the seventh line of'Y the target I6 is to be selected, a negative pulse is applied to the leads of the trigger circuits 85-86 and 9I-92. With current conduction in the left hand side of these two units, the application of seven pulsesto the lead 96 transfers current conduction to the right hand side of the trigger circuit 9I-92, producing a negative pulse at the lead 94 and a positive pulse at the lead 93. In selecting the second line of the target I8. a negative pulse is applied to the leads 95 o! the trigger circuits 81-88, 89-90 and 9I92. Simllarly, in selecting the sixth line of the target I9, a negative pulse is aonlied to the leads 95 of the trigger circuits 81-88 and 9I-92. Then pulses of a number determined by the value of u are applied to the terminal 96 of the circuit 85-86 until the set-up unit ls cleared and a positive pulse from the lead 93 of the last circuit 9I92 is applied to the horizontal deflection generator DH (Fig. 3) and a negative pulse is applied from the lead 94 of the circuit SII-92 to the switch SV and to the switch I2.

A battery 95 and reset switch 96 are provided for establishing a standby condition with current conduction in the right hand sides of the trigger circuit units IIS- 86, 81-88. 89-90 and 9I92.

The invention thus involves the provision oi.'

means 1) for deriving the products of a number of coefficients multiplied by different powers of one number, and (2) for deriving the sum of these products and another number.

I claim as my invention:

1. The combination of a member having horizontal rows upon each of which is recorded a different set of values oi' a first number and coemcients of different powers of a second number, means for forming an electron beam, means responsive to pulses representative of a third number for moving said beam to select one oi' said rows, means for moving said beam to scan said selected row and means responsiveto said beam during said scanning for combining said first number with each of said coeilicients multiplied by a diierent power of said second number.

2. In a computer for generating a function of a plurality of variables including ilrst and second numbers, the combination of a member having horizontal rows upon each of which is recorded a different set of values determined by said first `number and coeillcients of different powers oi' ll s said selected value with said selected coeiiicients multiplied by diiierent powers oi said second number. g

3. In a computer for generating a function of two variables including nrst and second numbers, thecombination of a member having horizontal rows upon each of which is recorded a dierent set of values determined by said first number and coeiiicients of different powers of said second number determined by said iirst number. means s for forming an electron beam, means responsive to pulses representative of said first number for moving said beam to select one of said rows, means for moving said beam to scan said selected row, means including a set-up unit wherein said second number is established for combining said selected value and coeiilcients multiplied by dif'- ferent powers of said second number,'and means for transmitting the scanned values to said oombining means. l

4. In a computer for generating a function of two variables including first and second numbers, the combination of a member having horizontal rows upon each of which is recorded a different set of values determined by said rst number and coemcients of different powers of said second 1 number determined by said first number, means for forming an electron beam, means responsive to pulses representative of said rst number for moving said beam to select one of said rows, means for moving said beam to scan said selected row, means including a set-up unit wherein said second number is established, a totalizer and transfer means connected between said second number set-up unit and said totalizer for cornbining said selected value and coemcients multi 12 plied by dinerent powers of said second number. and means for transmitting the scanned values to said totalizer. l

5. In a computer for generating a function ot two variables. including iirst and second numbers, the combination of a member having horiwherein said second number is established. a

totalizer and transfer means connected between said second number set-up unit and said totalizer, means for transmitting the scanned values to said totalizer, and means connected between said totalizer and said transfer means for combining said third number with said coeiilcients multiplied by different powers of said second number.

6. The combination of a member having horizontal rows upon each of which is recorded a different set of values of a rst number and coemcients of different powers of a second number, means for selecting said rows, means i'or moving said beam to scan one part of the selected row at a relatively high speed and .another part at a relatively low speed and means responsive to said beam during the relatively low speed part of said scanning for combining said first number with each oi said coeilicients multiplied by a diierent power of said second number.

GEORGE A. MORTON.

Images