DE1126653B - Device for the multiplication of non-binary numbers for a calculator working according to the series method - Google Patents

Description

GERMAN

PATENT OFFICE

'' ■: , i

D 24534 IX c / 42 m

REGISTRATION DATE: DECEMBER 19, 1956

NOTICE
THE REGISTRATION
AND ISSUE OF
EDITORIAL ·. MARCH 29, 1962

The invention relates to a device for multiplying non-binary numbers one by one Computing device working in series.

It is known to use a multiplication circuit in which the digit value of a digit of the one operand specifies how often the impulses that represent the other operand are forwarded to be counted on a counting device. The result of the multiplication is in read from a decadic counter. The pulses are generated by a continuously working pulse generator Taken via a gate circuit.

In practice, however, this method is limited to certain applications, since the Multiplication of multi-digit factors the required computing time becomes too long. It has already been suggested have been to use a circuit variant in which the arithmetic operations run partially in parallel. However, this requires additional circuitry for the parallel calculation.

It is also known to make use of multipliers or 1 × 1 circuits in the construction of calculating machines for the multiplication of which are built into the machine. It is possible to have partial products in the form of left-hand or right-hand To take digit components from special tables and to combine them by adding units. The partial products are taken from the tables provided in the machine. It's also in Known in another context to form sub-products anew each time. It is also known to use to work differently structured partial products that come about because the multiplicand is multiplied by all digits 1 to 9, so that the resulting results are partial products represent. In the multiplication method provided here, however, such partial products no use made.

Rather, the invention is concerned with simplifying and accelerating the calculation process a computing device that works according to the serial process, which with the help of cyclically rotating erasable memories works, in which the numbers to be offset are memory tracks and the places the numbers are assigned to successive sections of these tracks, and in which a device is provided during the supply of the two digits of the multiplicand to be offset and of the multiplier, a partial product is individually newly formed and in the form of a partial product corresponding to the partial product Pulse sequence is delivered to an adder. According to the invention, this is the adding device to multiply non-binary numbers for a using the series method

working computing device

Applicant:

Dr. Gerhard Dirks,
Frankfurt / M., Mörfelder Landstr. 44

Claimed priority:

Great Britain dated December 19, 1955 and February 25, 1956 (No. 36 266 and No. 5937)

Dr. Gerhard Dirks, Frankfurt / M.,
has been named as the inventor

The transfer memory allocated to the factory is designed to accommodate transfers beyond "1", and the adder is for receiving pulses corresponding to the partial products, as well as from the impulses corresponding to the carryover and of further, the significance of the previous results assigned and removed from the erasable memory pulses and only gives after execution this multiple addition the determined value to the cyclically circulating erasable Memory. This procedure has the advantage that the numbers can be entered without any major transformation, directly charged and can also be quickly withdrawn in an immediately accessible form.

The pulse series can e.g. B. consist of pulses that correspond to the digits 1 to 9 directly or also Code combinations correspond to 1, 2, 3, 5 or 1, 2, 4, 7.

A repeat switch can repeat the pulse trains corresponding to the digit value of the one Operands correspond, depending on the digit value of the relevant currently effective place value control of the other operand.

Furthermore, transmission devices are provided which have a storage capacity for more than one digit value, e.g. B. a storage capacity up to one Digit value of "8" or "9".

The means of transmission that have a storage capacity for digit values of "1" or more than "1", can consist of electronic or magnetic counting chains or multi-stage counting tubes. she

209 557/226

through the magnetic heads 4, 5, 6, 7, 8, 9, 10, 11/1 to U / n, 12/1 to 12 / n, 13/1 to 13 / n, 14 and 15/1 to 15 / « sensed or recorded. Counting stages 16, 17, 18, 19,

5 20, 21 and 22 used. The facility is controlled by signals recorded on the surface of the storage drum 1 and shown in Fig. 2 are shown in more detail.

The surface of the storage drum 1 is in several

o rere parallel tracks 23, 24, 25, 26, 27, 28 and 29 are divided. Each of these orbits is in turn into individual ones Sectors 30/1 to 30 / n divided. Between sector 30 / n and sector 30/1 there is an unused one Sector 31. Each of Sectors 30/1 to 30 / n is in

can e.g. B. have one or more multi-stage tubes.

The transfer memory assigned to the adder for the acceptance of more than "1" Transfers can consist of surface parts of a recording device, the predetermined Surface parts of the same recording medium can be that is also used to the imported To record digit values or the resulting digit values of the computing device.

It can also be beneficial when repeated
Addition of pulses from the selected pulse trains
a separate time period for the carry over for
at least one place to be provided, with digit values, eg 0 to 8 or 0 to 9, divided by a previous 15 three subsectors 32/1 to 32/3. The threefold arithmetic operations are stored, into which the division of the sectors 30/1 to 30 / n corresponds to a counting device. three-cycle calculation method, as described later in the

The transfer of signals or the connection with FIG. 1 can optionally be described. The impulses which indicate a carry over into the counting numbers to be processed in the invoice, d. H. device also simultaneously with the delivery of 20 the two factors, which are multiplied sol signals to the counting device for one or more flour are stored in the storage lanes 29 and 25 operands to be multiplied take place, with one digit in each of the the signals for the respective values for on-one-sectors 30/1 to 30 / n, with the last digit of each other counts are offset against each other. Number in sub-sector 32/1 of sector 30/1, the In one embodiment of the invention, 25 penultimate digit in subsector 32/1 of the sector namely, when multiplying two digits, represents 30/1, etc. is recorded.

the one digit value has a corresponding number of For reasons of space, only one is shown in FIG

Time intervals and the other numerical value represents a drum circumference with four sectors 30/1 to 30 / n dat- pulse train, which is a corresponding number of. Such a drum can accordingly have pulses in each time interval, whereby a maximum of a four-digit number can be recorded, which can be counted in the selected pulse train over this means by the computing device

either two two-digit numbers or one three-digit and one single-digit number are multiplied by each other, because only then will the result not be greater than four-digit 35. However, the maximum computing capacity is only depends on the size of the drum circumference and can therefore be expanded as required.

In the individual storage lanes 25 and 29 of FIG. 2, the two factors "83" and "67" are considered to be closed

A distribution device is effected, 40 multiplying numbers are recorded. The result where certain moments of time or positions in one of the multiplication, namely "5561", are in the memory-movement cycle the pulse trains in question track 24 recorded, and again the identify the last digit of the result in subsector 32/1 switched on for counting will. of sector 30/1, the penultimate digit in the first

The memory 45 sub-sector of the second sector 30/2, etc., assigned to each digit of a number.

area or that for scanning this memory area in the memory lanes 23, 26, 27 and 28 are

required time has at least two sectors or control or synchronization signals recorded.

Sections on, of which one is in each case. The signal in path 23 represents the beginning of the

Pulse train is assigned, which represents the number, while the scope of the storage drum 1, while the signals rend the other given for the arithmetic process 50 in memory path 28 each the beginning of one of the sec-

if including the transfer formation to the dispensers 30/1 to 30 / n and the signals in the memory path

supply stands, while possibly a third 27 in each case the end of one of these sectors 30/1 to

Mark section for writing down the result in front of 30 / n. In the sub-sectors 32/2 of the

is seen. individual sectors 30/1 to 30 / n are on the track

With reference to FIGS. 1 to 9, an input 55 26 is now recorded with ten synchronization signals.

direction according to the preceding explanations, their purpose later in connection with

wrestling described in more detail. It shows in detail Fig. 1 to be described. In the sub-sectors

Fig. 1 is a block diagram of the said computation 32/3 of the individual sectors 30/1 to 30 / n are not

setup, signals recorded but find it while

Fig. 2 shows the development of the surface of the in Fig. 1 60 this time, during which these subsectors at

magnetic memory shown, the imaginary zero line 33 move past, Ie-

3 to 9 basic circuit diagrams of the arithmetic operations to be described later in FIG. 1

electronic components. instead of.

The device contains a memory and a signal

encoder a drum 1, the surface of which is magnetized 65 of Fig. 1 are described: is cash. It is attached to the shaft 2 and is then recorded by a recording device

driven by motor 3. The signals on the magnetizing memory tracks 25 and 29 are generated Surface recorded signals are drawn, which the two multipüzier-

a number of time intervals represented by the first place value, a total count which is equal to the product of the two places.

The selection of a pulse train can be done by switching means, e.g. B. under the action of a relative movement between a recording device and a discharge device and / or a relative movement

5 6

The factors represent, contact 34 will be closed of the memory path 25 recorded signals

and thus amplifier 35 is open, whereby the possibility is not amplified by amplifier 42. The in amplification

possibility is created that the signals amplified on memory path ker42 are transmitted via line 46

23 recorded initial impulse fed to the input of counting stage 9 by magnet and count head 14 sensed and amplified in the amplifier 35 5 these each starting from zero white

can be. After this pulse in the amplifier ter. Accordingly, counting level 19 is set to the

35 has been amplified, it is first set via the line count value of the number of pulses

36 is fed to the amplifier 37 and from there corresponds to which in the first subsector 32/1 of the into the counting stage 22, which is recorded by each of these Im- first sectors 30/1 on the memory path 25 is incremented by one counting position, d. i.e., io net and which are the last digit of a to multibei each revolution of storage drum 1 will represent placing number.

Counting stage 22 advanced by one digit. By counting stage 19 is on lines 19/1 to

Counting stage 22 are over the lines 38/1 to 19/9, according to their respective counting position, a

38 / "the relay levels 39/1 to 39 /" are controlled. This excites the control levels 64/1 to 64/9 and thus the Stages are shown in more detail in Fig. 8, while the counter 15 creates the possibility that one of the

stage 22 corresponds to that shown in FIG. The rain 18/1 to 18/9 when counting through the counting level 18

The purpose of relay stages 39/1 to 39/9 is to feed any impulses that occur to the line 64 α,

one of the various magnetic heads 11/1 to 11 / «, through magnetic head 6 from memory web 28

12/1 to 12 / «, 13/1 to 13 /« and 15/1 to 15 / «are sensed on an impulse which marks the beginning of the second to switch on the appropriate amplifier. 20 tors 30/1. This impulse was

The arrangement of these magnetic heads is reinforced 47 in Fig. 2 and fed to the flip-flop 48, shown in more detail. Magnetic head 4 is on the imaginary This flip-flop is thus in such a switching zero line 33 arranged, likewise the magnetic heads 6, brought into position that amplifier 49 is opened, 7, 8 and 14. The magnetic heads 11/1 to 11 / «, 12/1 d. i.e., pulses sensed by magnetic head 4 up to 12 / "and 15/1 to 15 /" are arranged so that 25 and 50 can be amplified by amplifiers one of these heads on the imaginary amplifier 49 is further amplified and via line Zero line 33 is located, while the other 51 are each fed to the counting stage 16. By a distance of one sector from each other have magnetic head 4 sensed pulses, which in the and through contacts, which through the relay stages first subsector 32/1 of the first sector 30/1 39/1 to 39/9 are actuated, one after the other to the 30 memory tracks 29 are recorded, represent the last corresponding amplifier are connected to represent the second digit to be multiplied, and can. The magnetic heads 13/1 to 13 / n are so on - counting stage 16 is assigned to the count value by them, that the magnetic head is 13/1 by the width that corresponds to the number of these pulses, of two subsectors 32/1 and 32/2 from the counter stage 16 is via lines 16/1 imagined zero line 33 is removed and that the individual 35 to 16/9, according to their respective counting position, nen magnetic heads 13/1 to 13 / «to each other a Ab- one of the control stages 68/1 to 68/9 excited and thus stand, which has a sector width of 30/1 to create the possibility that one of the 30 / «corresponds to. Magnetic head 9 is arranged so that lines 17/1 to 17/9 when counting through the counter to the imaginary zero line has a distance, the stage 17 occurring pulses of the line 68 α zugeeinem corresponds to subsectors 32/1 to 32/3, 40 leads.

while the magnetic heads 5 and 10 with a sense of the pulse sensed by amplifier 44, wel-

stood from two such sub-sectors 32/1 to 32/3 rather the end of sector 30/1

are arranged opposite the zero line 33. also fed to the flip-flop 48 and thus closes

After now through the first through the magnetic head the amplifier 49. The through the two amplifiers

14 sensed pulse counting stage 22 from its counting 45 43 and 50 amplified pulses which the two

digit "0" was switched to counting digit "1", last digits of the two factors to be multiplied

if relay stage 39/1 is energized and thus represents the motors, the two flip-flops are also activated

gnet heads 11/1, 12/1, 13/1 and 15/1 are fed to the trains 52 and 53, so that these are turned into such

connected amplifier. The switch position brought about that amplifier 54

stronger 35 delivered pulse will continue to open the flip 5 °, d. H. that signals generated by

Flop40 is supplied and brings this into such a gneting head 8 to be sensed by the storage path 26

Switch position that on line 41 of the amplifier to be amplified by amplifier 54 and over

42 is opened, d. H. that in this amplifier Im line 55 and amplifier 56 in counting stage 18

pulse can be amplified. The amplifier 42 can take a long time. Furthermore, these signals are from

pulses are supplied which are fed through magnetic head 55 amplifier 54 via line 57 to the input of the im-

11/1 sensed by memory web 25 and fed to the amplifier pulse multiplier stage 58. This impulse

ker 43 were reinforced. By magnetic head 11/1 multiplier stage is built up so that through each

First, the pulses from memory path 25 on line 57 are supplied to the pulse on line 59

sensed which are emitted in the first sub-sector 32/1 of the ten pulses. The amplifier 54

first sector 30/1 are recorded. After the 60 is designed so that it only amplifies impulses

Sector 30/1 has passed the magnetic head 11/1, will or will let through when both flip-flops 52 and 53

by magnetic head 7 switched to memory path 27, d. H. if neither of the two

drew pulse which is the end of sector 30/1 to be multiplied by factors zero.

identifies, sensed and transmitted by amplifier 44

strengthens. This pulse is transmitted via line 45 to the flip 65 26 recorded pulses during the second

Flop 40 is supplied and switches it into its off-third of the first sector 30/1 sensed. This means,

initial position back so that amplifier 42 again they are only sensed when the two

is blocked, d. i.e., those in the remaining sectors on counting levels 16 and 19 by the magnetic

heads 4 and 11/1 of the two memory tracks 29 and 25 sensed pulses are already set. The first or each of the pulses appearing on line 55 causes the two flip-flops 60 and 61 switched so that the two amplifiers 62 and 63 respectively let pulses pass or strengthen. Flip-flop 60 is in its starting position by a pulse which occurs on line 64 a switched back, d. i.e., amplifier 62 leaves so

is switched so that amplifier 78 is opened. This makes it possible to use the magnetic head *), which is offset from magnetic head 8 by the width of a subsector 32/1 to 32/3, 5 is now sensed by the second sub-sector 32/2 of the first sector 30/1 on the storage web 26 Pulses which are fed to the amplifier 78 via line 79, in this amplifier 78 to amplify and via line 80 to the entrance

Long pulses from line 59 pass until they are fed to counting stage 20 at Lei. Through each of these acts an output pulse occurs. This pulse pulse is counting stage 20 by one counting position then on when counting stage 18, which of Lei counts.

tion 55 via amplifier 56 which by magnetic head 8 After counting station 20 has been fed to its zero position by sensed pulses, counting occurs on output line 72 that the counting position of the previous 15th pulse occurs via line 81 Flip-set position of the counting stage 19 corresponds, flop 77 is fed and this is in its ie, the pulse on line 64 a occurs when the initial position switches back, so that amplifier 78 so many pulses had occurred on line 55, is closed, ie, it As in the first sector 30/1 of the memory path 25, pulses no longer arrive in counting stage 20, which are recorded on line 81. During the same period of time occurring pulses are simultaneously fed to the flip-flop 82 from line 59 via the amplifier 62 so that the same number of pulses is switched over and thus ten pulses are supplied times as the number of pulses on line amplifier 83 is opened. This makes it possible for 55 to have occurred. These pulse groups of ten each, the further pulses occurring on line 79 via pulses were fed via line 65 to counting stage 17 amplifier 83 and line 84 to amplifier 85 and each have to feed them up to their full 25. These signals counting capacity are counted by amplifier 85. Furthermore, these were fed to the magnetic head 13/1, through which they are recorded pulse groups of ten pulses each via amplifiers in the first subsector 32/1 of the first sector 63 and line 66 and via diode 67 of the counting 30/1 on the memory track 24 , level 20 fed. Amplifier 63, which is controlled by the recording in the first subsector is da flip-flop 61, remains open as long as possible for magnetic head 13/1 to receive a pulse from flip-flop 61 by the width via line 68 α of two subsectors 32/1 to 32/3 opposite which is fed, to which this is offset in its initial zero line 33 (FIG. 2). The through position switches back and thus the signals recorded by the amplifier 63 magnetic head 13/1 represent the closes. A result of the first partial multiplication of the last pulse then occurs on output line 68 ″ if the counting positions of the two counting digits of the first factor are equal to the last digit of the steps 16 and 17, i.e. represent the second factor via the amplifier. This is achieved in that 63 get as many pulses each through amplifier on line 79 a total of ten pulses, of 62 supplied tens pulse group as in the first one, which is fed to the counting stage 20, thirds of the sector 30/1 of the memory path 29 are up - How the complementary value is drawn in the counting stage 20. Via amplifier 63, the value set in counting 40 is fed to the full counting value of stage 20 as many pulses as there are counting stage 20, ie "10". The remainder correspond to the product of the last two digits of the two pulses of the tens appearing on line 79 by the factors to be multiplied. The first pulse occurring on line 66, signaler 85, is fed to the amplifier by means of a pulse group via amplifier 83. This remainder corresponds to the number stored in counter via line 69 to flip-flop 70 45 stage 20. After the magnet is activated, this flip-flop 70 is switched and the head 9 opens the tens pulse group from the second network amplifier 71 - Memory path 26 was sensed, has now shown nerüberträ, ie after every ten of the entire first sector 30/1 supplied to the imaginary zero counting stage 20 pulses occurs on line 50 line 33 moves past. This means that in An-72 a pulse on, via amplifier 71 the input end to the sensing of the signals through magnetic line 73 of the counting stage 21 are fed. These tens carry-over pulses are used in this counting stage
initially saved. After magnetic head 8
all ten pulses, which are amplified in the first sector 55 by amplifier 44 pulse, such as 30/1 of the memory track 26 are recorded, already described, the flip-flops 40 and 48 were fed, is by magnetic head 5, which around
two subsectors offset from magnetic head 6
is sensed, the initial pulse recorded in memory path 28 in the first sector 30/1 and through the 60 line 86 to the flip-flop 53, whereby this is amplified in greater 74. Via line 75, the switch is switched back to its starting position and the amplified pulse is fed to the flip-flop 70 and closed. Furthermore, the pulse delivered by Vertet back to its starting position, stronger 44, is fed via line 87 to the amplifier 21, which closes it and thus flip-flop 82, so that this pulse occurring in its extension on line 72 is not switched back to the initial position and a further Im counting stage 21 can be supplied. At the same time pulse delivery through amplifier 83 is terminated. If the pulse appearing on line 75 via conductor 85 is controlled by flip-flop 88, which device 76 is fed to flip-flop 77, which causes it to be driven by pulses sent by the two magnetic heads

head 9 sensed by magnetic head 7 of the end pulse recorded on the memory web 27 in sector 30/1 and amplified by amplifier 44. Of the

leads, which prevents further sensing of signals by the magnetic heads 4 or 11/1 to 11 / n will. At the same time this impulse passes over

14 or 15/1 to 15 / «are sensed, each time being switched to a different switch position, d. H., by the pulse sensed by the magnetic head 14 at the beginning of one revolution of the storage drum 1, which is fed to the flip-flop 88 via amplifier 37 and line 89, amplifier 35 is opened, while the pulse sensed by one of the magnetic heads 15/1 to 15 / «, which is transmitted by the amplifier 90 is amplified and fed to the flip-flop 88 via line 91, the amplifier 85 is closed again will.

The second sector 30/2 now comes into the sensing position. This is done first by magnetic head 6 the recorded in the first subsector 32/1 of the second sector 30/2 on the memory web 28 Initial signal sensed and amplified by amplifier 48. This signal is, as previously described Way, the flip-flop 48 supplied. At the same time it also reaches the via line 92 Flip-flop 93 and thereby switches it to such a position that pulses, which by magnetic head 10 sensed by the second sub-sector 32/2 of the first sector 30/1 of the storage web 26 and are fed to the two amplifiers 95 and 96 via line 94, amplified by amplifier 95 and passed on line 97 while these pulses do not pass through amplifier 96 can. The line 94 pulses are simultaneously supplied, which through the magnetic head 12/1 of the first sub-sector 32/1 of the second sector 30/2 on memory web 24 are sensed. These Pulses represent the result of a possibly preceding partial multiplication. The magnetic heads 9 and 12/1 to 12 / n are arranged in relation to one another in such a way that pulses generated by one of the magnetic heads 12/1 to 12 / "are sensed, in each case at the point in time of the line 94 are supplied in which through Magnetic head 10 no pulse is sensed, i.e. that is, those generated by one of the magnetic heads 12/1 to 12 / « Pulses arrive in a pulse gap between two pulses sensed by the magnetic head 10. By the pulses appearing on line 94, which are amplified by amplifier 95 and the line 97 are supplied, is now counting stage 21, in which the tens carries over the previous Partial multiplication are stored, counted up to their full count. After this is reached, a pulse occurs on line 98, which is fed to flip-flop 93. This will this flip-flop is switched back to its initial position and thus amplifier 95 is closed and Amplifier 96 open, i.e. that is, the further pulses sensed by magnetic head 12/1 and magnetic head 10 are now amplified not by amplifier 95, but by amplifier 96 and the line 99 supplied. From there they reach the counting stage 20 via line 80 and set it to one certain count value. This count corresponds to the sum of a possibly in the second sector 30/2 of the memory path 24 recorded result of a previous partial multiplication and the in the counting stage 21 stored number of tens carries.

This sum represents the starting position of the counting stage 20 for the following partial multiplication. In this second partial multiplication becomes the penultimate digit of the first factor which is in the subsector 32/1 of the second sector 32/2 is recorded on memory web 29, with the last digit of the second Factor which is recorded in the first subsector 32/1 of the first sector 30/1 of the memory web 25 is multiplied. The multiplication is done by the signal passing through the magnetic head 6 was sensed at the beginning of the second sector 30/2, flip-flop 48 in the manner described above switched and thus amplifier 49 was opened. This makes it possible to use the magnetic head 4

ίο now in subsector 32/1 of the second sector 30/2 of the memory track 29 sensed pulses to the counting stage 16 and thus to a To set the count value which corresponds to the count value of the penultimate digit of the first factor to be multiplied is equivalent to. Counting stage 16 was previously triggered by the end pulse of sector 30/1, which was generated by magnetic head 7 sensed and amplified by amplifier 44 is reset to its initial position. this happened in that the amplifier 101

ao this pulse was supplied and that this pulse via line 102 to the zero of the counting stage 16 was fed, whereby this was set to its zero position. The second factor of Partial multiplication, namely the last digit of the second factor, was already being used during the sensing of the first sector 30/1 in counting stage 19 is set. In the following, the previous one is repeated in context the process described with the first partial multiplication, d. i.e., counter 20 receives so many pulses are supplied, as correspond to the result of the second partial multiplication. Likewise be the pulses occurring on line 72, which represent tens carryovers, in the previously described Way stored in counting stage 21 and after the result of the second partial multiplication in the subsector 32/1 of the second sector 30/2 was recorded on memory web 24, transferred to counting stage 20, whereby the starting conditions for a third partial multiplication are created. Afterward follow as many partial multiplications as the first factor contains digits, i.e. i.e., the last digit of the second Factor becomes 1 with all digits of the first factor during one revolution of the memory multiplied. Now that the memory 1 has performed one revolution, the magnetic head 14 the initial pulse recorded in memory path 23 is sensed a second time and in the previously described Way of the counting stage 22 is supplied. This in turn advances this counting level by one digit and thus relay stage 29/3 energized, whereby the corresponding magnetic heads 11/2, 12/2, 13/2 and 15/2 can be connected to the associated amplifier. Accordingly, during the second revolution by magnetic head 11/2 the recording in the first subsector 32/1 of the second sector 30/2 sensed on the memory web 25, d. That is, counting stage 19 is now set to the value of the penultimate digit of the second factor is set. The sensed by magnetic head 14 and by amplifiers 35 and 37 amplified pulse is, in addition to counting stage 22, also the zero position of the counting stage via line 19/0 19 supplied. This will be based on the second digit of the second factor before the new hiring the initial value »0« is set.

The last, penultimate, etc. to to the first digit of the first factor and set counter 16 accordingly. While

209 557/226

11 12

the second revolution of the memory 1 will occur and therefore negative Imnach at point 120 each digit of the first factor with the penultimate pulse arise, soft via capacitor 121 of the Number of the second factor multiplied. The result output line 49/2 and from there via line 51 The results of these partial multiplications are in turn fed to the counter stage 16. Will now in the first subsectors of the individual sectors to 5 on input line 48/2 via capacitor 122 dem Memory track 24 recorded, namely the grating of the right system of the double triode 110 is a Result of the first partial multiplication, d. H. If the positive pulse is supplied, then this flip result flips multiplying the last digit of the first flop back to its original position, and at point Factor with the penultimate digit of the second factor, 113 occurs through the anode current, which is now on the first subsector of the second sector 30/2 io again via the right system of the double triode 110 the memory web 24 recorded, etc. The flowed on, a strong negative voltage drop, which recorded The value corresponds to the sum that the grid of the triode 116 is so far ahead of the negative the count value which during the first spans that positive pulses, which are recorded on input rotation in the relevant subsector, are fed to line 49/1, through triode 116, and the result of the partial multiplication of 15 are blocked.

last digit of the first factor with the penultimate one. The remaining controllable factors used in FIG

Digit of the second factor in previously described amplifiers and flip-flops are basically of

Way was formed. same structure as that shown in FIGS.

These partial multiplications are then switched into the should flip-flops by negative pulses different for all digits of the second factor, from 20 one and positive pulses in the other switch position starting at the back, executed, d. that is, the memory 1 are brought in, then the supplied ones are turned off for the complete implementation of a multi- tive impulse through a corresponding i? C element plication as many revolutions as the second factor differentiates and the positive part of the differential gate Owns. Correspondingly, the result of the counting stage is connected to the input of the flip-22 via a diode Equipped with as many counting levels as a maximum of 25 flops are supplied, thereby doing this in the same way digits are provided for the second factor. how it is switched by positive impulses.

After counting stage 22 to the last counting point in FIG. 5, the pulse multiplier stage 58 is in

has been counted further, the locking of the Kon is shown in greater detail. This consists of

clock 34 canceled and thus through amplifier 35 a self-excited multivibrator, which the dop-

no new start signal sensed. 30 contains peltriode 123, a controllable amplifier

With reference to FIGS. 3 to 9, the one with the triode 124 and a monostable flip-

individual electronic components used in Fig. 1 Flop, which contains the double triode 125 and the

shown in more detail. Amplifier 124 controls. How the stage works

With reference to FIGS. 3 and 4, the following is first of all: By the multivibrator, which by The flip-flop 48 interacts with the control 35, the double triode 123 and the corresponding switchable Amplifier49 explained. The starting position elements are formed, positive pulses are earth Flip-flops 48 is such that the right system of the testifies which, via capacitor 126, the grid of the Double triode 110 is live. This means triode 124 will be fed. This grid is over that through the common resistance and grid resistance 127 with the anode of the left side the anode resistor 112 a current flows, 40 stems of the double triode 125 connected. The cathode which at point 113 is a strong negative voltage - the double triode 125 via cathode resistance voltage drop compared to earth. This span 128 is connected to negative pole 129 while the leg drop is connected to the anodes of the double triode 125 by the two via grid resistor 114 Grid 115 of the triode 116 of the amplifier 49 pulling anode resistors 130 and 131 as well as the common leads. The grid 115 is thus connected to the series resistor 132 at ground 45 with ground potential Cathode 117 is so strongly negatively biased that they are. The left grid of the double triode 125 is that pulses from input line 49/1 via resistor 133 and the common pre-capacitor 118 are fed to this grid, the resistance 132 also connected to ground potential response threshold the triode 116 does not reach the. By flowing grid current will be able to exceed the left or, i. That is, they are biased by 50 systems of the double triode 125 so that the triode 116 blocked. If this system is now connected to an anode 48/1 the flip-flop 48 a positive pulse pulling current flows and consequently through the anode resistor leads, then this pulse passes through capacitor 130, a negative voltage drop occurs. This 119 to the grid of the left system of the double triode negative voltage drop is via grid lead-110. As a result, the grid bias on this 55 resistor 127 is applied to the grid of the triode 124 via the grid increased so far that a tilting process occurs, causing this to be negatively biased so far and the left system of the double triode 110 is current- that positive pulses, which via capacitor becomes a leader. At the same time, the right sy- 126 will be fed to this grid, the responses the double triode 110 blocked. Point 113 is the threshold of triode 124 not reached or above it can only exceed the voltage drop over 60 and thus in this triode 124 stood 111 negative to earth. The grid front is not reinforced, but locked. Is now tension at grid 115 of triode 116 there is thus briefly more from input line 57 via capacitor 134 set below the threshold of this triode. In the grid of the right system of the double triode pulse, which is now supplied with a positive pulse via input line 49/1 to 124, then the are fed to the grid 115 of the triode 116, 65 right system of the double triode 125 current-carrying, The grid voltage of this triode is now able to. As a result, a negative anode resistor 131 occurs to change between the response value and zero, voltage jump, which occurs via capacitor 135 whereby a corresponding anode current swan is transmitted to the grid of the left system,

thereby locking this left system. The duration of the blocking state depends on the time constant of resistor 133 and capacitor 135 . During this blocking time, no current flows through the anode resistor 130 , so that the grid of the triode 124 is only biased negatively with respect to ground by the voltage drop across the common series resistor 132. This voltage drop is such that positive by the triode 124 pulses which are fed via capacitor 126 to the grid of the triode 124, 124 are amplified and thus negative pulses occur at the anode of the triode. These negative pulses are fed to output line 59 via capacitor 136. The frequency of the multivibrator and the time constant of the resistor 133 and capacitor 135 are dimensioned so that during the opening time of the amplifier 124, i.e. as long as the left system of the double triode 125 is blocked, ten pulses from the multivibrator via capacitor 126 to the grid of the triode 124 supplied and amplified by triode 124. The total duration of the ten pulses amplified by triode 124 is somewhat shorter than the pulse spacing of the pulses supplied via line 57.

6 and 7 show the interaction of the counting stage 16 and one of the control stages 68/1 to 68/9. The counter stage 16 contains the glow counter tube 137, to which negative pulses are fed from input line 51 via capacitor 138. As a result of these negative pulses, the glow discharge, which exists between anode 139 and one of the cathodes 140/0 to 140/9 , is switched on from the respective cathode to the next. If there is a glow discharge between one of the cathodes 140/0 to 140/9 and the anode, a current flows through the associated cathode resistor 141/0 to 141/9 , which causes a positive voltage drop across this cathode resistor. This positive voltage drop across the cathode resistors 141/1 to 141/9 is used to control the corresponding control stages 68/1 to 68/9 .

In Fig. 7, the control stage 68/3 is shown in greater detail. It contains the triode 142, the grid 143 of which is connected via resistor 144 to point 145 of the voltage divider, which is formed from resistors 146, 147 and cathode resistor 141/3 . The voltage divider is located between negative pole 148 and earth and gives the grid 143 such a negative bias voltage in the idle state that positive pulses, which are fed from line 17/3 via capacitor 149 to grid 143, are blocked by triode 142. If there is a glow discharge between the cathode 140/3 and the anode 139 , then such a positive voltage drop occurs across the cathode resistor 141/3 that the voltage at point 145 is increased to such an extent that pulses are supplied via line 17/3 , now amplified by triode 142 and fed via capacitor 150 to the common output line 68a of all control stages 68/1 to 68/9.

Negative pulses can be fed to the cathode 140/0 via line 102 and capacitor 151 , as a result of which the glow discharge migrates from any point to this cathode and thus the counter tube 137 can be set to zero. The counting stages 19 and 22 are basically of the same circuit structure, with only line 102 being omitted in the case of counting stage 22 . Counting stage 21 also only differs from the one shown here in that the control lines 16/1 to 16/9 are no longer used, while line 102 corresponds to line 98 at counting stage 21.

The counting stages 17 and 20 differ from the other counting stages in that their maximum counting frequency is at least ten times as high as that of the counting stages 16, 18, 19 and 22, while the maximum counting frequency of the counting stage 21 must be at least twice the value of these other counting stages . For example, if the pulse repetition frequency of the recordings on memory 1 is 15 kHz, then counting stages 16, 18, 19 and 22 must have a maximum counting capacity which is greater than 15 Hz, while counting stage 21 must have a maximum counting capacity greater than 30 kHz . This necessity arises from the fact that pulses which are sensed by one of the magnetic heads 12/1 to 12 / n are each in a gap between two pulses which are sensed by the magnetic head 10. Accordingly, pulse trains with twice the pulse frequency than the recordings on memory 1 appear on line 94. The ten times the counting capacity, ie greater than 150 kHz, of counting stages 17 and 20 is due to the fact that ten pulses are triggered in pulse multiplier stage 58 by each pulse appearing on line 57 and fed to the two counting stages 17 and 19. While counting tubes of the Dekatron type can be used for counting stages 16, 18, 19 and 22 , for counting stage 21, for example, a counting tube of the ElT type and in the counting tubes 17 and 20 a counting tube such as the "coaxial trochotron" of the RYG 10 type be used.

8 shows one of the relay stages 39/1 to 39/1, which are controlled by the counter stage 22. The control takes place via the corresponding control line 38/1 to 38 / «, which are connected to the individual counting cathodes in place of the control lines 16/1 to 16/9 shown in FIG. 6. The relay stage contains the triode 151, the cathode of which is grounded, while the grid of this triode is biased by the voltage divider, which is formed from the resistors 152 and 153 , via resistor 154 in the idle state so that only such an anode current through the triode 151 flows, which is not sufficient to energize the relay 155, which is switched on in the anode line of the triode 151 . If the voltage on the corresponding control line 38/1 to 38 / n is increased due to the existence of a glow discharge on the corresponding cathode, the grid bias voltage on the triode 151 also increases, so that the anode current increases and the corresponding relay 155 is excited . When the relay is energized, four contacts are closed, one of which is assigned to one of the magnetic heads 11/1 to 11 / ", 12/1 to 12 / n, 13/1 to 13 /" and 15/1 to 15 / ". In this way, one of these magnetic heads is connected to the associated amplifier in accordance with the position of the counter tube 22.

Fig. 9 shows the controllable record amplifier 85 in greater detail. This amplifier contains the triode 156, the grid 157 of which is supplied from line 84 via capacitor 158 with positive pulses. The triode 156 is formed by the flip-flop 88 in that shown in connection with FIGS

Controlled in a manner, that is to say, depending on the switching position of the hip-hop 88, there is a more or less negative voltage on the control line 159. This negative voltage is transmitted to grid 157 via resistor 160. If the flip-hop 88 is in such a switch position that control line 159 is strongly negative with respect to ground, then the grid bias voltage on grid 157 is so strongly negative with respect to the cathode of triode 156 that pulses are sent to grid 157 via capacitor 158 get through the triode 156 are blocked. If, however, the hip-hop 88 is switched to the other switch position, then pulses which are supplied via the capacitor 158 are amplified by the triode 156 and, after being amplified via the capacitor 161, are supplied to the grid 162 of the triode 163. At the same time as the triode 156 , the triode 164 can also be controlled by a change in potential on the control line 159. The potential at grid 165 of triode 164 is first determined by the two resistors 166 and 167 ao in conjunction with the voltage applied to control line 159. The resistors 166 and 167 are dimensioned so that the anode current of the tube 164 is blocked when a strongly negative voltage is present on the control line 159.

If the control voltage on control line 159 is made less negative, which corresponds to opening of triode 156 , then the grid potential on grid 165 rises and reaches approximately the value "0", so that the corresponding signal heads 13/1 to 13 / " , which is connected to the cathode 168 of the triode 164 , a relatively large current flows. This current is used to erase signals which are recorded on the memory web 24 of the memory 1 moving past the magnetic heads 13/1 to 13 / n.

The grid 162 of the tube 163 is so strongly negatively biased via the grid bleeder resistor 169 that practically no anode current flows through the tube 163 in the idle state. This means that the current through the corresponding one of the magnetic heads 13/1 to 13 / n is determined exclusively by the current through the tube 164 . However, if positive pulses are fed to the grid 162 of the triode 163 via the capacitor 161 , the grid bias voltage of the tube 163 moves between the response value and the value "0", so that an anode current flows via the resistor 170. This anode current initially causes a voltage drop across resistor 170, which has an effect on grid 165 of triode 164 via resistors 171 and 172.

Since the voltage at point 173 is kept relatively stable by capacitor 174 , namely at the potential "0", the voltage drop across resistor 170 is communicated to the grid of triode 164 according to the resistance ratio of resistors 171 and 172 . Since the voltage drop across resistor 170 is negative with respect to the cathode 168 , the grid 165 is also negative with respect to the cathode, as a result of which the anode current through the triode 164 is reduced. Since the sum of the currents through the tubes 163 and 164 flows through the corresponding one of the signal heads 13/1 to 13 / «, when the tube 163 is driven by positive pulses, the direction of the current is reversed and thus pulses are recorded by the corresponding one Signal heads 13/1 to 13 / «on the storage path 24 of the storage drum 1. This reversal of the current direction is caused by the fact that the voltage drop across the resistor 170 becomes so large, when the tubes 163 are adjusted accordingly, that the anode current through the tube 164 is blocked . In this case, only the anode current of the tube 163 flows through the corresponding one of the signal heads 13/1 to 13 / n. However, this current flows from ground to a negative potential, while the current flows through tube 164 from a positive potential to ground. Since the control of the tube 163 is done by short pulses, the magnetization of the magnetic layer of the storage web 24 of the storage drum 1 is effected for a short moment in the opposite direction than during the erasure, which corresponds to a recording of a signal.

Claims (6)

PATENT CLAIMS: 1. Device for the multiplication of non-binary numbers for an arithmetic unit working according to the serial method with the help of cyclically circulating extinguishing devices in which the numbers to be offset are assigned memory tracks and the digits of the numbers are assigned to successive sections of these tracks and in which a device is provided in which During the supply of the two digits of the multiplicand and the multiplier to be calculated, a partial product is newly formed individually and sent to an adder in the form of a pulse sequence corresponding to the partial product, characterized in that the carry memory assigned to the adder for the acceptance of more than "1" Transfer is designed and that the adder is designed to receive pulses corresponding to the partial products, further pulses corresponding to the carry and other pulses assigned to the values of the previous results and taken from the erasable memory sen is formed and only after this multiple addition has been carried out does the determined value transfer to the cyclically circulating erasable memory. 2. Device according to claim 1, characterized in that between the count values of the Multiplicand time segments are provided in which the serial adder multiple counts which are selected depending on a position of the multiplier. 3. Device according to claims 1 and 2, characterized in that the count of the multiplicand is represented by equally spaced counting value pulses, the number of which is equal to the Count value of the relevant position is, and that the pulse frequency for the multiple count values is selected is that optionally controllable by the multiplier up to nine multiple pulses the series adder can be supplied, the number of times multiple pulses are delivered in one place the number of count value pulses is determined. 4. Device according to claims 1 to 3, characterized in that for the count value carry a device is provided which, in addition to the "0", notes at least eight count values can that arise when counting the multiple pulses into the series adder as the sum of the tens of carryovers during the passage of a Position. 5. Device according to claims 1 to 4, characterized in that in a special working cycle when processing the next following digit, the count value carry is transferred to the Senenadierwerk. 6. Device according to claims 1 to 5, characterized in that the count value carry is stored in a special memory path. Considered publications:
German Patent No. 896 571;
U.S. Patent No. 2,609,143;
British Patent No. 731140; "Handbook of Industrial Electronics", Berlin, 1954, p. 137; "High Speed Computing Devices," McGraw Hill Book Comp., Inc., New York-London, 1950, p. 65 to 73, 299 to 301; "Arithmetic Operations in Digital Computers", D. van Nostrand Comp., New York, 1955, p. 151 to 161, 247 to 271, 326; "Electronic Engineering", October 1953, pp. 407 to 410; "Synthesis of Electronic Computing and Control Circuits", Cambridge Massachusetts, Harvard University Press, 1951, pp. 200 to 201, 207; "Electronic Rundschau", No. 5, 1955, pp. 196 to 202; "Valvo, Technical Information for Industry", 5.4.1955, pp. 1 to 18. For this purpose 2 sheets of drawings © 209 557/226 3.62