DE1096084B - Computing device - Google Patents

Description

The invention relates to electrical computing devices in which digits and numerical values are in coded or non-coded form can be represented by pulses. With the device z. B. Additions and subtractions and with a corresponding extension of the control also multiplications and Divisions are carried out.

The invention particularly relates to an electrical computing device in which pulse trains that Represent decimal places of a number, are supplied as standard to the formation of totals and counters in which the result is stored on a rotating magnetic memory.

According to the invention, a pulse train is fed to a counter with such a time delay compared to the other pulse train that originally simultaneous pulses arrive at the counter one after the other, and the head recording the result is offset by one recording length of the decimal number with respect to the sensing head of the path of one summand, so that the result is used at the same recording location from which the one summand was taken at the beginning of the calculation.

The pulse series are not fed to the counter one after the other, but in such a way that the pulses alternate one row with the other row, d. H. that the impulses of the two impulse trains are nested. This measure has compared to known institutions in which the Pulse series are fed one after the other to a counter, the advantage that the counter or the arithmetic unit with double the pulse repetition rate, d. H. can work at increased speed.

The difference in the timing of the pulse trains is chosen so that the pulses of one pulse train with those of any other pulse train do not coincide in time, so that a single counter the Record nested pulses and form the result from a number of pulse trains can.

All pulse trains can be derived from the same source; you can z. B. by a mutual relative movement can be obtained between a pulse generator and sensing means.

The pulse trains can run continuously and only be made effective for the duration of the count. However, you can also only have a predetermined number of pulses each from the to counting digit value depends.

According to a further feature of the invention, the various pulse trains are through a corresponding Number of mutually offset sensing means generated, with between the sensing means and one or more signal generators a relative rectifier

Applicant:

Dr. Gerhard Dirks,

Frankfurt / M., Mörfeider Landstr. 44

Claimed priority:
Great Britain 25 February 1956

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

movement takes place. The various pulse trains can also be offset from one another by a number of Signal generators are caused, between which and a corresponding number of sensing means a relative movement takes place.

The signal generator can be designed as inductive generators that z. B. permanent magnets on a non-magnetic carrier material, or they can contain interruptions in a homogeneous magnetic material or also contain teeth or strips on a wheel or disk or finally consist of recorded signals.

In another embodiment of the invention, optical generators are used for signal generation used, the z. B. optical markings with different reflectivities or different Transparency included.

The signal generators can also be designed as capacitive generators with corresponding capacitive sensing means cooperate. So z. B. Conductors are used whose potential differs from that of the environment. They can also be stabilized by a zero potential or some other Shielded potential.

In other cases, a contact generator can also be used as a signal generator.

If there is a relative movement in the signal generator, the movement between the pulse generator and the sensing means can be synchronized or run as a directly coupled movement. Signals that control the relative movement can be derived from a relative movement between the

009 680/272

3 4

sensing means and the recording medium derived from the number are interrupted. It is also possible,

will. optionally the impulse output in certain time

There can also be a single impulse generator to start the moment when a new body is opened

are applied, whose impulses z. B. comes with the help of.

Rectifiers, multivibrators or the like. Doubled 5 In the former case, the delivery of pulse or be multiplied. trains interrupted or from an impulse train

According to a further feature of the invention switched to the other at a time that

the different pulse trains can be determined by time-fixed signals that different

Time slot generated with the help of a cathode ray indicate numerical values. These timed

that is repeated across different screens

is distracted. cyclically working memory, which is not a circulating

In one embodiment of the invention, magnetic memories will need to be removed.

the pulse trains are fed to e.g. gate circuits, the signals can be taken from a memory,

are controlled by signals, which the storage elements to be changed in horizontal and / or

represent calculating digits. The gate circuits 15 vertical connections arranged crosswise

can be tube circuits or magnetic connections.

Orders included. The control signals have a The memory elements can consist of magnetic cores

predetermined timing and will consist of sensing or electron tubes.

means of an erasable memory, e.g. B. a magnet The timely sampling of this memory can

drum, removed. 20 with the help of electronically or magnetically operated

The memory is preferably effected as a cyclical work-distributors, which the concerned

the magnetic memory is formed. The different elements or groups of elements in one another

Digit values can make the following operations effective through different lengths.

Recording areas by different times The pulse trains can according to the invention also

lying signal sections or by various 25 can be derived from pulse trains, which are based on the

Be marked with pulse numbers. appropriate places of the number assigned area-

The starting point in time for the effectiveness of a sharing of a record carrier, e.g. B. a ma-pulse train can through the end of a number of gnetisbaren circulating memory, recorded recurring signal recordings on the record are, the recorded pulses in each of a recording medium or through the beginning of a 30-digit assigned area marks the signal series to be processed be. It can also display the numeric value. The point in time that represents the numerical values is selected as the starting point in time; B. a number of sensing devices may be offset, and / or recorded sinusoidal oscillations, on which the sensing devices themselves can have a counter-record carrier expired or have expired staggered arrangement,
According to the invention it is possible to use the interfering signals present in the computation, e.g. B. to allow the noise process to take place in two work cycles, to be able to distinguish between signals. This oscillation whereby in the one work cycle the numerical values represented by the pulse signals, which trains by their presence or absence 40 are processed, z. B. act, can have a predetermined frequency. can be added to get a result while

The starting point in time for the effectiveness of one in the second work cycle which shows the result-

Pulse train can also be identified by a pulse from the counting stage to the memory or

be drawn, to which a predetermined amplitude and another device are fed, wherein

Has direction, or it will be supplied by the change 45 if necessary, a carry signal

determined by the recording direction or by one.

upward or downward frequency jump of the

frequency recorded by the sensing means, a certain number of pulses are added to the counting stage

or phase. be guided. Reaching or exceeding

The aforementioned timed signals 50 of a predetermined value, e.g. B. a limit value, control z. B. electronic or magnetic switch, then triggers a signal that the beginning or the which in turn influence the gates. The end of the delivery of pulse trains causes the Pulse trains offset from one another become the pulse result of a recording device, a counter supplied to tubes as flip-flop chains from the pointing device or other consumer can be built up or also from 55 feed.

Transistors or magnetic elements, e.g. B. The signal which, upon reaching or over-coring, Toroidal cores and the like, consist or even from stepping the predetermined value is triggered, capacitive elements are built. As an elec- tric switch, z. B. a bistable electron tubes Will usually operate multi-stage tubes ironic or magnetic switches that used. 60 the delivery or interruption of pulses or

It may be advantageous to use the counting device to record two other signals, the display

To supply pulse trains, one of which controls the od.

has twice the frequency as the other. This switch is preferably an electronic one

Devices can also be provided with or magnetic double-throw switch or a deadden the pulses of the basic frequency the numerical values 65 circuit, which the switching paths for two trains of the one number to be calculated, during signals, namely a signal train for the recorder Pulse train of double frequency is used for the voltage and the other for the cancellation of the energy, specify the second operand. drawing controls. In such a case, the

The pulse output by the pulse trains can be converted into the signal train that causes the recording, a Si

predetermined time moments corresponding to the points 70 signal generator are taken, the signals before-

specific phase and frequency, so that a second recording takes place in exactly the same places like the previous recording of similar signals. Such a signal generator can also be a Main record of signals included.

The signal generator can also be permanent signal generating devices, e.g. B. permanent magnets, as the main signal generator or pulse generator.

The other signal train for the erasure can be an erase generator, e.g. B. a high frequency generator, can be removed.

When the signals are recorded for a resulting digit value at the second operating cycle are to be, then they are preferably assigned to the number in question in the same place Written down the area from which the digit value signals of one operand were taken, while the first work cycle of the calculation process was carried out in the relevant location.

The switch for controlling the switching paths for the two signal trains can extend the time the recording time of the signals or a shortening of the recorded signal paths with the help cause of cancellation signals; the instant in time or transition point from the recording surface of the signals for the erasure surface of the signals can correspond to the point in time of the result signal, which results from the offsetting of the two operands of the relevant position.

This point in time can be achieved by reaching or exceeding a predetermined value within the counting level can be determined.

The invention also relates to the fact that the counting stage, e.g. B. a multi-stage tube, a counting in two directions, d. H. in a positive and a negative direction. This creates a Subtraction possible in which the signal pulses of one operand in the opposite direction are counted.

During the subtraction, the fixed-time signals of one of the two digit values in the as described above, as in the case of direct addition, while the signals for the other digit value are used in a complementary manner. This effect is achieved that in the addition process the pulses from the starting point (zero point) one Instead, up to the point in time of the digit value signal, are effective while for the subtraction process the impulses are so effective that complementary numerical values up to "nine" are added, wherein the counting pulses begin and their delivery at the instant of the actual digit signal is ended at time "9" so that the complementary value is displayed.

The output of a signal that represents a calculation result can be controlled by that the counting stage pulses are supplied, which they during the second work cycle of a place in in the opposite direction, so that the zero crossing marks the instant of the result signal supplies.

If the predetermined limit is exceeded during the first work cycle, then speak Carry out devices to effect a carry over to the next digit. These transfer devices can be an electronic or magnetic relay or recording medium for storage of the carry signal included.

The actuation of the transfer device causes that during processing and preferably on Start of processing of the next digit a carry pulse is delivered to the counting device will.

This carry pulse is generated as a fixed-time signal from a signal generator, e.g. B. from one recorded or a permanent signal on a movable recording medium, and is output to a gate circuit controlled by the carry signal from the previous digit and the passage or blocking of the carry pulse to the counting device depends on it makes that the gate has previously been operated or not, depending on whether a The carry signal from the calculation process of the previous digit was present or not.

The carry pulse can upon a return of the carry device to the initial position in a defined moment of time are given or not given, preferably at the beginning the processing of the next digit, the delivery of the carry pulse to the counting stage thereof depends on whether the carry device in processing the digit value signals of the preceding Position has been actuated.

According to the invention, the computing process can take place in two working cycles, one working cycle for processing, e.g. B. Addition of the numerical values, serves and the inclusion of a carry-over includes the previous digit and / or the issue of a carry signal for the next digit, while the second work cycle records the result of the calculation. The calculation process can be carried out in each point can also be carried out in three work cycles, namely

a) the carry over, z. B. a single or multi-digit carryover, e.g. B. for a multiplication,

b) the processing, e.g. B. the addition of the numerical values, and

c) the recording of the result.

A counting chain or counting tube, e.g. B. a multi-stage tube, can for any mehrziff erige carry formation, z. B. be provided for transfers from zero to eight in a multiplication from one place to the next.

The calculation process can have separate stages for the processing of two-digit values with intermediate storage and can also have a separate step for deleting the previous result contain.

The process between recording and billing will usually take place in series, however, it can also take place in parallel.

The working steps of sensing and recording can be carried out alternately; it can there is also a gap between the record in one place and that in the next be. The gaps can be used to make a recording during a further work process to undertake, e.g. B. for further digits of a number.

According to the invention, two or more sets of counting stages can also be used in successive processes can be used in combination with the recording means. In any case, this includes the position assigned area of the recording device usually also an area which is not for the Recording is used, but a transition time is taken into account during processing. Of the

7 8

Area of a location can be, for. B. ten areas- Fig. 14 shows the principle of a device for

parts for recording and two surface parts for offsetting binary numbers.

show the transition period. The previously described method of operation is initially

The devices for dividing larger time on the basis of the block diagram, which in Fig. 1 gesignalen or pulses in smaller time signals or 5 shows, explained in more detail.

Pulses can contain cathode ray tubes, the device shown there contains the drum 1, have a number of screens, which when their cylindrical surface with a magnetizing-scanning the dependent ble layer is provided by the cathode ray. This surface is in the control periodic signals, e.g. B. to generate individual tracks 2 to 6 divided. The lanes 2 multiple signal trains for multiplication. The io and 3 represent memory paths in which the one cathode ray tube a deflection can be represented in individual numerical values to be calculated, at least one deflection direction or also a deflection, but there could also be other erasable memories, have steering over a circular path; they can be the sum through on these storage lanes 2 and 3 Control electrodes are controlled. The screens lead through an input device, not shown Cathode ray tubes can be recorded to control the device. Areas 4, 5 and 6 Output signals are used, the input and / or carry records to generate pulses. Output devices one or more multi-stage These records can either by Än-tubes may contain. changes in the magnetic state of the surface

In one embodiment, the rotating drum 1 is generated according to the earth. At

finding a decadic counting stage with a counting 20 their position can also be permanent magnets, which

capacity of ten pulses used and after that in a non-magnetizable body of the drum 1

Series principle worked by, at the lowest let in, kicking.

Beginning digit, digit by digit, add the summands. The individual tracks 2 to 6 are to be added along. This offsetting of a place, ie the start of the drum 1 in individual sectors 7/1 to 7 / n · the partial addition takes place, as mentioned above, divided into 25, of which one for each place of two or more work cycles. During the first number to be offset is provided. In this second working cycle, the pulses, which are gates 7/1 to 7 / n, are in turn divided into two sub-summands, sectors 8/1-w and 9/1-n scanned by the memory. Added within and in the counting stage. In the following ar- sub- sectors 8/1 to SIn are on the two memory clocks the stored in the counting stage 2 and 3 the numerical values of the individual digits result of the partial addition are again recorded on the memory on the two numbers to be calculated, and the counting level is reset to zero. This recording takes place in the present case that the starting position for the following partial game is given by oscillations of different lengths, such as addition. this can be done, for example, in subsector 8/1

The results can be recorded on the 35 licher. As can be seen from the figure the same location from which it was before is in the first part of subsector 8/1 in the one of the summands was sensed. These two storage tracks 2 and 3 each have a low summand can in turn record the result of an earlier frequency oscillation 10 or 11, be executed computing process, which in contrast during the second part of subsector 8/1 Memory was recorded. The new one, added by 40 through a high-frequency oscillation 12 or 13 of the second summand is erased or one erased by high frequency The result is then permanently displayed in the same storage location. The distance of the frequency hop held and in turn serves as a summand for a 14 or 15 forms from the end of the subsector 8/1 further addition. thereby the characteristic for the size of the

Since the execution of a partial addition and the 45 count value to be offset in this position

draw the resulting result to the digits.

If there are different times, there are no undesirable within subsectors 8/1 to 8 / n Couplings between the recording on the two areas 5 and 6 pulse groups and sensing devices, so that special sensors 16/1 to 16 / n and 17/1 to 17 / "each with" 9 "screen means can be dispensed with. 50 pulses recorded. The impulses of the impulse group

By adding up in places and 16/1-n are arranged so that they are opposite Immediate recording of the results of the partial the pulses of the pulse groups 17 / l-? i by one In addition, the counting stage only needs to be offset by half a pulse interval for a decade. Both impulse designed to be what a substantial savings groups will be thanks to two magnetic heads 18 and 19 respectively results. 55 is sensed and processed by the amplifiers 20 and 21

The invention is not based on the decadal strengths. You will be using the two controllable gates

leadership, but can also be fed to the counting stage 24 at other 22 or 23. How many im-

Type of counting can be used as well as with Dar- pulse of each of the pulse groups 16/1 to 16 / n resp.

position of the digits by code pulses different from 17/1 to 17 / «are fed to the counting stage 24 is

Art. 60 depending on the length of the high-frequency

The description includes Figs. 1 to 13, which drawing 12 or 13 or the length of the erase

show in detail the following: area in the corresponding subsector 8/1

Fig. 1 shows as a block diagram a simple one to 8 / ″ on the two memory tracks 2 and 3. These

direction to illustrate the working method; Recordings on storage lanes 2 and 3

FIGS. 2 to 10 show basic circuit diagrams of the circuit diagrams in FIG. 1 65 are used by the two magnetic heads 25 and 26 Block pictures; felt and through the two amplifiers 27 and 28

11 shows the block diagram of an addition amplified. From there the signals are sent to the two

and subtracter; Control stages 29 and 30 supplied. Through these two

FIGS. 12 and 13 show basic circuit diagrams of the two gates 22 in control stages 29 and 30

Fig. 11 block images used; 70 and 23 affected. The control of these gates 22 and

23 takes place in such a way that pulses which are sensed by the two signal paths 5 and 6 last so long the gates 22 and 23 can not pass, as by the two magnetic heads 25 and 26 the low frequency Recording is sensed by the two storage lanes 2 and 3. After the frequency jump 14 or 15 has passed the magnetic heads 25 and 26 and the erasure areas 12 and 13 now through the magnetic heads 25 and 26 are sensed, tilt the two control stages 29 and 30 and thus open the two gates 22 and 23. This has the effect that pulses generated by the two magnetic heads 18 and 19 of the two signal paths 5 and 6 can now be fed to the counting stage 24. The switching of the two control stages 29 and 30 depends on whether the corresponding control stage a low-frequency vibration corresponding to the records 10 and 11 is supplied or not. The frequency of these oscillations is such that about five to six periods correspond to the pulse spacing correspond to the recording on the signal paths 5 and 6, respectively. This frequency ratio was chosen in order to make any interference impulses or background noises ineffective through integration. The control stages 29 and 30 are constructed in such a way that when remain of more than three oscillations of the recordings 10 or 11, these control stages in their Tilt back the starting position and thereby switch the two gates 22 and 23. The two amplifiers 27 and 28 are constructed so that only the low frequency Vibrations 10 and 11 are amplified, while the high-frequency extinguishing areas 12 and 13 are not amplified, so that the control stages 29 and 30 are not influenced by this either can.

While now within a sub-sector 8/1 to Bin by one of the records 16/1 to 16 / n and one of the records 17/1 to YJIn , pulses are offset from one another, the number of which corresponds to the sum of the two count values stored in the memory lanes 2 and 3 are recorded in the corresponding subsector 8/1 to Bin , are fed to the counting stage 24, a pulse is generated at the counting stage 24 if this sum of the two counts is greater than "9". This tens carry pulse is fed to the memory stage 32 via line 31. The memory stage 32 contains a flip-flop, which is switched over to another switch position by the pulse fed to line 31; as a result, this pulse is stored in the memory stage 32 until the counting stage 24 is ready again for the calculation of the next digit, ie until the result stored in the counting stage 24 is recorded in one of the memory lanes 2 or 3. Following this recording, a pulse is fed to the storage stage 32 via line 33, which flips the flip-flop of the storage stage 32 back into its starting position, whereby a pulse is generated which is fed to the counting stage 24 via line 34 and returns to the initial position "1". brings. The pulse on line 33 occurs when the second subsector 9/1 -n of the corresponding sector 7/1-w has moved past the magnetic heads. As a result, the pulse 35/1-w recorded in signal path 4 at the end of the corresponding subsector 9/1-n is sensed by magnetic head 36 and amplified in amplifier 37. This pulse is now fed from this amplifier to the storage stage 32 via line 33.

Before this pulse occurs on line 33 and thus also on line 34, counting stage 24 must be brought to their zero position. This is done by the fact that the synchronizing stage 38 on the Line 39 of the counting stage 24 pulses are fed. The synchronization stage 38, which is shown in more detail in FIG is shown, contains two signal generators, which consist of the two signal paths 40 and 41 and the two magnetic heads 42 and 43 are formed. The signal tracks 40 and 41 represent peripheral tracks a rotating body 44, which is either directly coupled to the rotating drum 1, running synchronously or identical. The signal paths 40 and 41 are also in individual sectors 7/1-ra and these in turn are divided into individual subsectors 8/1-M and 9/1-w. This division corresponds to that of the tracks 2 to 6 on the rotating drum 1. On the signal track 40 are within each subsection recorded 9/1-n pulse groups with "10" pulses each. The representation of these impulses can, as already described in connection with the signal paths 5 and 6, through recordings on a magnetizable layer or by permanent magnets attached to an insulating body are, or are caused by other local changes in the magnetic state. While the sub-sectors 9/1-w are in the sensing position, magnetic head 42 ten pulses each sensed and amplified by amplifier 45. These impulses are transmitted via line 39 fed to the counting stage 24 and continue to count them by one counting position. At the beginning of everyone Subsector 9/1-w is recorded in signal path 41 for each »1« pulse 46/1-w, which is generated by magnetic head 43 is sensed and amplified by amplifier 47. Each of these pulses is transmitted via line 48 of the Control stage 49 supplied, whereby this is switched so that low-frequency signals which generated by generator 50 and fed via line 51 to the input of control stage 49, this happen and can thus be fed to the magnetic head 52 via line 51. The magnetic head 52 is so arranged on the circumference of the drum 1 that it is when the magnetic head 26 at the beginning of one of the Subsectors 9/1-w of storage lane 2 is at the beginning of the previous subsector 8/1-w of the same Speicherbahn is located. Since magnetic head 43 (Fig. 2) is arranged at the same height as magnetic head 26, this means that control stage 49 is switched over just at the point in time when magnetic head 52 is at the beginning of the corresponding subsector 8/1-w.

The recording of the low-frequency oscillation by the magnetic head 52 now continues until the control stage 49 is switched back to its starting position. This takes place in that when the full counting capacity "10" is reached, the counting stage 24 emits a pulse which is fed to the control stages 49 and 54 via line 53. This pulse switches control stage 49 so that signals from generator 50 are blocked, while control stage 54 is switched so that the high frequency generated by generator 55 is now fed via line 56 to magnetic head 52 and thus to the remainder of the corresponding subsector 8/1 to 8 / w is awarded. The length of this high-frequency recording on the corresponding subsector 8/1 to Bin now represents the result of the previous addition. This is achieved by the fact that after counting stage 24 during the passage through the corresponding subsector 8/1 to Bin through the sensing position of

009 680/272

1 090084

Magnetic heads 25 and 26 on the sum of the two counts, which are in the memory lanes 2 and 3 in the corresponding subsector 8/1 to 8 / w are recorded, plus any existing one Tens carry is set and that after passing this corresponding subsector 8/1 to 8 / w during the time in which the following subsector 9/1 to 9 / w is in the sensing position at the Magnetic heads 25 and 26 are located, "10" pulses from pulse generator 38 via line 39 of counting stage 24 are fed. These "10" pulses are now counted by the counting stage 24 from the result of the previous one Beginning with partial addition, continue to its full counting capacity »10«. Upon reaching this Counting capacity occurs the aforementioned pulse on line

53, and there is thus a frequency jump in the recording, which is caused by magnetic head 52. The high-frequency recording that now follows lasts as long as the corresponding subsector 8/1 to 8 / w is in the recording position with respect to magnetic head 52. The interruption at the end of the corresponding subsector 8/1 to 8 / m is caused by the corresponding one of the pulses 35/1 to 35 / w, which are recorded in signal path 4. This takes place in that, after the corresponding subsector 8/1 to Bin has passed the magnetic head, the associated subsector 9/1 to 9 / w has also moved past the magnetic head 36. As already described, this magnetic head 36 senses the associated pulses 35/1 to 35 / "at the end of the corresponding subsector 9/1 to 9 / w and supplies it to amplifier 37. From there this pulse arrives via line 33 to the storage stage 32 and triggers a pulse which is fed via line 34 to the counting stage 24, which is now in its zero position, if the result of the previously performed calculation was greater than "9". The pulse amplified in amplifier 37, however, also reaches the control stage via line 57

54 and switches them over so that the high-frequency signals generated by generator 55 are blocked and thus no further recording is made by the magnetic head 52.

The function sequence just described takes place during each of the sectors 7/1 to 7In, ie two numbers to be calculated are added one after the other digit by digit, since each of the sectors 7/1 to 7In corresponds to one digit. This makes it possible to calculate numbers with any number of digits with a single-digit counting level, since the maximum capacity of the computing device only depends on the number of individual sectors 7/1 to 7In on the circumference of drum 1.

The individual electronic components used in FIG. 1 will now be described in FIGS. 3 to 10.

FIG. 3 initially shows a basic circuit diagram of the two amplifiers 20 and 21. These are normal low-frequency amplifiers, which have the two double triodes 58 and 59 as amplifying elements contain. The operation of the amplifier is such that signals transmitted through the corresponding magnetic head 18 and 19, respectively, are fed to the control grid of the left half of the double triode 58 and thus be amplified in this triode. Via capacitor 60, the amplified signals are dem right system of the double triode 58 is fed and further amplified there. From the anode of the right System of the double triode 58, they reach the left grid of the second double triode via capacitor 61 59 and from the left anode of this tube via capacitor 62 to the right grid, so that they are in the right system of this double triode 59 are reinforced again. They come from the right anode Via capacitor 63 to the output line and are fed from there to the two control stages 22 and 23, respectively. The amplifier 37 basically has the same structure, however, in this amplifier the signals which are taken from the right anode of the double triode 58, not just via the capacitor 63 is fed to line 57, but also to line 33 via capacitor 64 and from there to of storage stage 32.

The two amplifiers 27 and 28 differ from the above only in that their Gain decreases sharply with increasing frequency. They are designed so that only the low frequency Record 10 and 11, respectively, on storage lanes 2 and 3, generated by generator 50 is amplified and the high frequency recording produced by generator 55 is disabled. this is achieved in that between the right anode of the first double triode and the left grid In the second double triode, a π-element is built in as a frequency-determined component, which is made up of the two Capacitors 65 and 66 and the inductor 67 is formed. The cutoff frequency of this τι-member is like this dimensioned that it is higher than the audio frequency generated by generator 50, and the attenuation curve is such that the high frequency generated by generator 55 is blocked. Accordingly, the Magnetic heads 25 and 26, although both records from the corresponding subsectors 8/1-w on the Memory tracks 2 and 3 sensed, the output lines, which to the control stages 29 and 30 lead, but only the audio-frequency signals 10 and 11 are supplied. The grid bias of the right The system of the second double triode of these amplifiers 27 and 28 is set so that the output lines only the positive half-waves of the sinusoidal recordings 10 and 11 are supplied will.

In Fig. 5, one of the control stages 29 or 30 is shown in principle. This tax bracket is paid via the The positive half-waves of the recordings 10 and 11 are fed to the input line and thus load the capacitor 70 on the grid of the triode 71 to a certain value. This grid 72 of the double triode 71 is biased so strongly negative with respect to the cathode 74 via resistor 73 that the Anode current in the idle state is almost zero. This means that if on the input line no positive signals are supplied, at point 75 the voltage prevails, which is determined by the voltage divider, which is formed from the two resistors 76 and 77 is generated at point 78, since on the anode resistor 79 only has a very small voltage drop due to the negligibly low anode current is produced. If, on the other hand, positive signals are fed to the input line, then capacitor 70 is charged to a positive voltage, which increases the voltage on grid 72 increased compared to the cathode and thus a larger anode current can flow. This generates anode current at anode resistor 79 a negative voltage drop, so that point 75 is strongly negative compared to Earth will. This negative bias is applied via line 80 to the corresponding one of the gates 22 and 23, respectively fed. These gates 22 and 23 are designed so that they are in the idle state, i. H. when the

13 14

speaking control stage 29 or 30 no positive left system of the double triode 91 supplied, then

Signals are supplied, are open and thus the flip-flop switches over so that now the left

Pulses, which through the corresponding the Ver system of the double triode 91 is current-carrying and

stronger 20 or 21 pulls the input of these gates - thus leading to a negative voltage jump at point 93, let pass. The corresponding 5 occurs. The pulse on line 31 is then delivered

of the control stages 29 or 30, then occurs when counting stage 24 is counting the

device 80 the mentioned negative voltage drop, stronger 20 or 21 amplified pulses, which over

and the corresponding one of the gates 22 and 23 will be fed to the gates 22 and 23 of the counting stage 24

closed and thus no longer lets impulses through. the one that reaches its maximum counting capacity, d. h., the Line 81 establishes the connection of the earth potentials io. The pulse supplied to line 31 represents a tens

between the two levels. transfer.

In Fig. 6 one of the aforementioned gates 22 or at the end of a computing period, during which each -23 is shown in principle. It contains the triode 82, because a point of two numbers to be added grid 83 is connected to the net via grid bleeder resistor 84, a positive line 80 is now connected to line 33. The cathode 85 is connected to a 15 pulse, which is caused by the corresponding on-ground potential and is generated via line 81 with the drawing 35/1 to 35 / n . This associated control stage 29 or 30 is connected. If this positive pulse is not energized via capacitor 94 to the associated control stage, then the right grid of double triode 91 and this grid 83 of triode 82 switches only such a large negative back to its starting position. As a result, voltage is applied as blocked by the voltage divider 20, the left system of the double triode 91, and at the point 93 formed from the two resistors 76 and 77, a positive voltage jump occurs. is generated. This positive voltage jump, like this negative bias voltage, is so great that the preceding negative voltage is differentiated by capacitor 95 triode 82 only slightly below its response value and resistor 96. In the case of the previously negative bias, ie positive pulses, 25 mentioned negative voltage jump, a negative pulse occurs on input line 86 via capacitor 87 differentiation result, which is fed to grid 83, grid can be blocked by diode 97. With the bias voltage now taking place between the response value and the change in the positive voltage jump, however, the value zero changes and a positive pulse is thus amplified in the triode differentiation result, which 82 because the diode 97 can pass through the grid voltage fluctuation 30 and the voltage via capacitor 98 caused anode current fluctuations is fed to output line 34. From there, the anode resistor 88 receives negative voltage fluctuations and this positive pulse goes to the counter stage 24 and sets conditions which are fed to the output line 90 via the capacitor 89 to the output value "1" for the next. If the billable body is a.

associated control stage 29 or 30, however, then 35 Flip-flop that controls a triode. The flip-flop biased strongly negative with respect to the cathode 85 contains the double triode 99, while the controllable one is. This negative bias voltage is so great that the amplifier stage is formed by the triode 100 , positive pulses which are fed to the input line 86 of the 4 "The circuit and mode of operation of the flip-flop grid 83, this grid 83 must not be assumed to be known Anodes over the response value of the tube 82 are controlled by the double triode 99 are over the two anodes, so that no anode current fluctuation resistances 100 and 101 and the common occurs and the Im-series resistor 102 supplied at this point in time connected to ground potential, while pulses are blocked control stages 29 and 30, 45 rend the cathode of the double triode 99 through resistor Kathodennur in the presence of the high frequency up 103 is connected to -15OVoIt. the drawings 12 and 13, or in the absence cathode of triode 100 is above magnetic head 52 of a record are not energized , are thus also at ground potential, while the anode is through this the gates 22 and 23 only impulses which triode are recorded via anode resistor 104 with + 150VoIt on the signal tracks 5 and 6, 50 is connected. The grid 105 of the triode 100 is allowed to pass through when the high-frequency recordings 12 or 13 from the anode of the right system of the double triode 99 are sensed through the magnetic heads 25 or grid discharge resistor 106 with point 107. When there is no connection. This makes it possible to bias the grid 105 drawing on the memory lanes 2 or 3 to different degrees negatively, even through the magnetic heads 18 and 19 no pulses 55 are sensed as the starting position of the flip-flop, since at this time each take that the right system of the double triode 99 subsectors 9/1 to 9 / w is carrying the corresponding magnet current. Accordingly, there is a head pass at point 107 and no negative voltage drop in these subsectors, which is present due to the span recordings. voltage drop across the common resistor Fig. 7 shows the principle circuit diagram of the memory level 60 102 and in addition, through the anode current through 32. This memory stage by a flip-flop, the right system overall the double triode forms resulting 99 containing the double triode 91st The circuit shown by the negative voltage drop across the anode resistance, which is formed assuming 101 is known. If this strong negative voltage is allowed, the double triode 91 can drop into two stable states via grid bleeder resistor 106 , ie it can alternately transmit 65 grids 105 , whereby signals which are current-carrying to the left or the right system. Input line 51 are supplied and via condensers-The initial position is such that the right system sator 108 reach the grid 105 , the grid front double triode 91 is live. If the voltage of this grid cannot increase via the response value storage stage 32 from line 31 via condenser triode 100 and thus a positive pulse to the grid of tube 100 is blocked by capacitor 92 and not amplified. Will

109 ^ 084

IS 16

on the other hand, the flip-flop is switched over, so that the left-hand side. The counting stage 24 contains as a system is blocked, then at point 107 there is only an essential component of the ten-stage counter tube 120, a negative voltage drop, which is continued to be counted by pulses from one counting position to the voltage drop at the common series resistor 5. These pulses, which are fed to 102 and the voltage drop across the anode resistors, lines 34, 121, 122 and 39, stand 101, which is converted into a divider by the cross current of the voltage through a pulse shaper stage, which consists of the two resistors 109 and for counting purposes. This Im-110 is formed, arises. The pulse shaper stage at point 107 thus contains the double triode 123. When the prevailing negative bias voltage, which reaches the grid 105 when the counting capacity "10" of the counting stage 120 is reached, a negative pulse occurs at the auxiliary anode 124, so great that the tube 100 to its normal, which causes the monostable multivibrator, which sets the double operating point and thereby contains signals, triode 125, to tilt, so that further pulses are taken from input line 51 via capacitor 108 of this stage to grid 105 , an anode current change 15 can. At point 126 this monostable multivibrator in the tube 100 can cause. A positive pulse occurs at the anode, which is supplied to either line 31 or line current flowing from earth potential via the winding of 53 via the current of this tube or the equivalent cathode double triode 127. This control stage, which by magnetic head 52 to the cathode. By changing the double triode 127 is controlled, this current is accordingly a recording 20 by the flip-flop with the double triode 128.
of the signals fed to line 51 on the upper surface of the counter stage opposite to the magnetic head 52 on the input lines 121 and 122, negative pulses from the two gate memory tracks are possible. The control of the flip-flop stages 22 and 23 are supplied. These negative impulses are effected by positive impulses which are sent from the input via diode 129, capacitor 130 and line 48 and input line 53 of this stage pulling resistor 131 to the left grid of the double triode. The control stage 123 is switched on. This grid is connected to the cathode 133 via the resistors 131 and by pulses soft from the input lines 48 132, so that the or 53 a via capacitor 111 is the left grid of the grid bias voltage OVoIt. The right grid double triode 99 can be fed. Due to this positive double triode 123, on the other hand, lies above its grating pulse, the left system of the double triode 30 becomes conductive resistor 134 at the junction of the 99, while the right system blocks resistor 135 with resistor 136. Accordingly, the voltage increases at point resistors 135 and 136 form together with 107, and the tube 100 can be controlled by signals which are sent to resistor 137 to a voltage divider to which line 51 is fed. The cathode voltage of the double triode 123 is distributed in a positive manner on the lines 53 and 57, respectively. The right grid is supplied to the pulse, then this passes via capacitor double triode 123 is thus biased via resistor 134 so 112 to the right grid of double triode 99 and far negative that in the right system, the flip-flop switches over so that the right system in the rest position of this flip-flop, practically none of this double triode 99 is energized. This causes anode current to flow. Due to the negative pulses, point 107 becomes strongly negative, and the pulses fed to the left grid of double triode 123 on line 51 40 cannot change the anode current, will cause the anode current of the left generation in tube 100, or the anode current system accordingly the pulse shape is weakened so this tube is zero. that the negative voltage drop across anode resistors-Fig. 9 shows one of the generators 50 or 55, stand 138, which is in the idle state, briefly generating the signals that are reduced by the 45. This positive voltage corresponding to the control stages 49 and 54 is controlled via capacitor 139 to the right grid. This is a normal double triode 123 and thus the grid oscillator, which contains the triode 113, whose bias voltage of this system corresponds to the pulse anode circuit through the resonant circuit, which increases briefly from form. The resulting anode capacitor 114 and the inductance 115 formed 50 current increase lasts until the charge is matched. Via the inductance 116 of the capacitor 139, the new voltage ratio is followed by a feedback to the grid of the triode nits and the right grid through 117, so that self-excitation occurs. The voltage drop across the cathode resistor 135 via the frequency-determining parts of the oscillator, i.e. grid leakage resistor 134, in turn, a negative capacitor 114 and inductance 115 as well as the bias voltage, so that the right system is blocked again if frequency-dependent feedback by the coil and the left system again 116 and capacitor 118, are energized accordingly. Dimensioned at the same time as the frequency to be generated. When the gene current increases in the right system of the double rator 50, these parts are dimensioned in such a way that a triode 123 increases the cathode current and low-frequency oscillation occurs, which causes a positive voltage jump across capacitor 119 from the anode of triode 113 . This positive voltage can be picked up and via line 51 the jump is fed as a positive pulse via capacitor control stage 49. The generator 55 is constructed 140 of the left deflection electrode 141 of the counter tube 120 in such a way that a high-frequency oscillation is supplied. The counter tube 120 is a commercially produced, which is controlled by the control stage 54 65 is a common design of the type ElT, which we will. Both vibrations can be assumed to be known to the magnetic head. 52 supplied, that is, through this magnetic head, by supplying positive pulses to the steering electrode 141 according to the position of the two control stages 49, the electron beam, which and 54 either emits a high-frequency or a low-frequency one from the cathode of the counter tube 120, is recorded in each case with a frequency take place. 70 deflected one step to the left and took

17 18

after this diversion again a new stable gene addition was equal to or greater than "10". The state a. In the tenth deflection step, the water pulse on line 31 occurs when the electron beam hits the auxiliary anode 124, so that the flip-flop formed by the double triode 125 through resistance 142 between the positive pole via anode 142, which a pulse supplied via capacitor 143 to auxiliary anode 124 who have favourited the anode-cathode route and 5 flips. Furthermore, the occurrence of a pulse on the cathode resistor 143 is an anode current line 31 from the position of the flip-flop, which flows. This current caused at the anode resistance is formed by the double triode 128, depending. 142 a voltage drop, which as a negative im- This flip-flop is actuated by the negative pulses pulse via capacitor 143 to the left grid 144, which is fed to lines 121 and 122 or double triode 125. He represents the sign io 39 to be fed. Lines 121 and 122 show that the full counting capacity of the counter tube has negative pulses during the first phase 120 and that this counter tube now has to be added at every point, that is, it must be supplied to the time pulse which the elec- point at which the sub- sectors 8/1 to SIn tronenstrahl and thus the counter tube 120 brings itself into its sensing position with respect to the magnetic heads 18 catch position. The via capacitor 143 15 and 19 are located. These pulses are transmitted via line pulse can still be used for other control 158 and capacitor 159 in the left grid of the purposes. The resetting of the electric double triode 128 is supplied, whereby the grating pre-electron beam in the counter tube 120 takes place, voltage on this grid briefly so low that the double triode formed by 125 is monorized that the left system of this double triode stable flip-flop, which in its How it works and 20 128 is blocked. By means of feedback, the circuit structure of the input stage, which increases the grid bias of the right system, formed by the double triode 123, corresponds to the fact that the right system of the double triode 128 conducts current through the capacitor 143 of the left grid. The negative pulse supplied to 122 is reversed by the first negative pulse supplied to line 121 or 144, and thus a brief anode current flow through the flip-flop is switched from its initial position to the other anode resistance 145 and a brief increase in the stable switch position. Because of this, the cathodic current takes place. By switching the current flow, the potentials at the points through the anode resistor 145 change via this 160 and 161. These potentials are, however, a determining negative voltage drop, which is more negative for the position of the control stage, which is caused by the pulse via capacitor 146 to the grid 148 of the 30 double triode 127 is formed. This is caused by triode 149 being guided. As a result of this negative Im- achieves that the grid 162 is blocked for a short time via grid leakage pulse, so that the anode current is zero at 163 with point 160 and grid 164 via grid and the voltage leakage resistor 165 is connected to point 161. In drop across anode resistor 150 disappears. Since the initial position of the flip-flop 128 prevails, the anode of the diode 151, the potential 35 point 161, has such a strong negative bias that the operating voltage is applied and thus positive positive impulses which are sent via capacitor 161 to the opposite of their associated cathode . This has reached grid 164, the grid bias voltage of this connected to the right deflection electrode 152. The opening of the diode 151, the electrons migrate and thus the anode current through the right beam remains blocked in its starting position in its starting position in a relatively short time. On back, that is, the counter tube 120 is set to the count value point 160, on the other hand, there is only a negative zero to the extent that it is. The short reset time is therefore tension that the grid 162 to just below its desired, since it is biased the distance between two mutually responsive values. Positive pulses, subsequent pulses and thus the counting frequency, which this grid 162 agrees to via capacitor 167. Since the anode current interruption of the tube 45 is carried out, the grid preload 149 can therefore only take place briefly, the anode of the diode will change between the response value and the value 151 after the potential increase through the voltage zero, so that these pulses cause a voltage drop the anode resistance 150 immediately brought the anode current change in the left system of the double back to a potential which is caused by the lower triode 127.

This anode current change is generated at the left electrode 152. In this case, the anode of the anode of the double triode 127 is supplied with negative pulses, and the diode 151 is supplied with negative pulses compared to the associated cathode via the capacitor 168 of the output line 53 and thus the resistance of this diode 151 will be very high. If, on the other hand, the flip-flop is switched from its initial position in relation to the resistance value of the resistance value to the other switch position 153. As a result, there is no impairment 55, then the voltage at point 160 becomes strong in the counting process. negative, while the voltage at point 161 assumes the negative value in the same way as via lines 121 and less strongly. Thus 122 the left system of the double triode 123 negative now positive pulses, which are supplied via capacitor pulses, this is done via the 166 supplied to the grid 164, the anode line 39. The negative pulses 60 supplied there current in the right system of the double triode 127 get through diode 156 and capacitor 130 at heights, and thereby entd the left grid of the double triode 123 at the right anode. During the described, in each case by one counting point further, and the first calculation period, that is, while the subsectors and the reset functions run in the same 65 8/1 to SIn in the sensing position on the magnet way. From line 34 is located via diode 157 heads 18 and 19, respectively, the control is then fed a negative pulse when the stage with the double triode 127 is set so that Imbefore are amplified by a tens carry pulse on line pulse to grid 164 and via line 31 the memory stage 32 has been switched over and thus 169 can be fed to the line 31, ie it was established that the result of the preceding 70 if during this time in which the addition

of the two numerical values to be calculated takes place, a pulse is fed to the grid 144 via the capacitor 143 is, then the resulting positive pulse at point 126 on line 31 generates a negative Pulse. The storage stage 32 is then supplied with a pulse if during the first Computing section the counting capacity of the counting stage 120 is reached. During the second calculation period negative pulses are fed to the counting stage 24 on line 39, which the counting tube 120 of the The count value previously obtained by the addition continues to count up to its full counting capacity.

As soon as this is reached, a negative pulse occurs again at auxiliary anode 124, which likewise reaches the flip-flop via capacitor 143, which is formed by the double triode 125, and in the manner described above, the counter tube 120 in its Resets the starting position. At the same time, the positive one created by overturning the flip-flop The pulse at point 126 is fed to the two grids 162 and 164 via the two capacitors 166 and 167. By the first pulse fed to line 39, which via capacitor 170 to the right Grid of the double triode 128 is supplied, the flip-flop formed by this double triode is in its Switched back the initial position and thereby blocked the right system of the double triode 127 while the left system amplifies pulses and feeds them to line 53 via capacitor 168. This is it is possible to use the data at point 126 during the second calculation period, i. H. while subsectors 9/1 up to 9 / rc pass the magnetic heads 18 or 19, pass the positive pulse on to line 53, where it can be removed for further tax purposes, d. i.e., he's used to the two To switch control stages 49 and 54, as previously described.

With the computing device just described, it is possible to add or expand the Control also perform subtractions. In connection with this exemplary embodiment, a further described, which also allows additions and subtractions to be carried out. at A further development of the above-mentioned calculation principle was applied to this exemplary embodiment, which allows you to work with less effort. This second embodiment is based on Hand of the block diagram, which is shown in Fig. 11, described. The one shown in Fig. 11 appears to be greater effort compared to the block diagram in Fig. 1 is due to the fact that in this second embodiment, the corresponding subtraction device was also taken into account. That In contrast to FIG. 1, the block diagram in FIG. 11 was only drawn single-pole in order to improve clarity to ensure the representation.

In this second example, another embodiment according to the invention is to be described in more detail, namely that the representation of the individual numerical values to be calculated is not done by a frequency recording of a defined length, but rather that they are represented by individual pulses in a corresponding coding. This means that in this embodiment the number of pulses 16/1 to 16 / w or 17/1 to 17 / «which are fed to the counting stage 24 are not determined by frequency recordings, as in FIG. 1, but instead of the Ploch - and low-frequency recordings 12 or 13 are in this version pulse recordings 171/1 to 171 / M ./. 1 or 172 to 174/1 to 174 / w ./.!, which determine the count of the corresponding digit. This means that the memory path 2 and signal path 5 as well as memory path 3 and signal path 6 have each been replaced by a signal path 175 or 176/1 to 176 / m. The signal paths 175 or 176/1 to 176 / m are also here sections of the cylindrical surface of a rotating storage drum 177. This storage drum corresponds to the drum 1 in FIG , be replaced. Just like this, it is divided into individual lanes 175, 176/1 to 176 / w and 178 to 185. The tracks 178 to 184 are used for synchronization or control purposes, while the memory tracks 176/1 to 176 / w are used to record the calculation result or a summand to be added. The second summand to be added is recorded on memory track 175, while the complementary value to the count value recorded on memory track 175 is stored on memory track 185. This is the complementary value to the number "9", which is required for subtraction purposes. In this embodiment, too, the circumference of the rotating drum 177 is divided into individual sectors 186/1 to 186 / n . Each of these sectors 186/1 to 186 / n is in turn divided into two subsectors 187/1 to 187 / w and 188/1 to 188 / w, respectively. While sub-sectors 188/1 to 188 / κ are passing the signal heads, a number to be calculated is added or subtracted, but when sub-sectors 187/1 to 187 / w are passed, the result of a previous calculation is written out, i.e. The writing process takes place when the subsectors 187/1 to 187 / w move past the magnetic heads 189 to 198. The result is recorded, however, by the magnetic heads 199/1 to 199 / w, which are offset from the associated sensing heads 191/1 to 191 / w by the width of a subsector 187 or 188 in the direction of rotation 200. Thus, the result is recorded in each of the sub-sectors 188/1 to 188 / n during the time during which the sub-sectors 187/1 to 187 / w are in the sensing position with respect to the magnetic heads 189 to 198. In this embodiment, too, a single-digit counter stage 201 serves as the counter through which the individual digits of a number are added or subtracted one after the other. The counting stage 201 is counted further by pulses which are controlled by the controllable gate stages 202 and 203 and amplified by the amplifiers 204 and 205. During additions, amplifiers 204 are supplied with pulses which are sensed by magnetic head 191 from memory web 175.

These pulses sensed by the magnetic head 191 pass through the switching position α of the contact 206 to the input line 207 of the amplifier 204 and are connected to the controllable gate stage 203 via line 208 fed.

From the controllable gate 203, these pulses are transmitted via line 209 to the input of the controllable gate 202 supplied. This controllable gate 202 is also supplied with pulses via line 209, which sensed by one of the magnetic heads 193/1 to 193 / w and amplified by amplifier 205. the Pulses which reach the controllable gate 202 via the controllable gate 203 represent a value to be added Summands, while the pulses, which are supplied via line 209, represent the second summand or an already existing result for which

further numbers are to be added. If a number is to be subtracted from another or from an already existing result, then switch 206 is switched to switch position b so that now pulses which are sensed by magnetic head 192 from memory path 185, via amplifier 204 and the controllable gate 203 to the controllable gate 202 arrive. These pulses, which are recorded on signal path 185, each represent the complementary value to the number "9" of the digits recorded on path 175. The recording on these two memory paths 175 and 185 is carried out by a known input device not shown here. The just mentioned pulses reach the counting stage 201 via the controllable gate 202 and line 210. The input of these pulses in the counting stage 201 is controlled as follows: By closing the contact 211 by a control or similar device, the controllable gate 212 is opened, see above that a pulse which is sensed by magnetic head 197 from signal path 183 and amplified by amplifier 213 can pass this gate.

Only one pulse 214 is recorded on path 183, namely at the end of subsector 187 / « of the last sector 186 / m. This pulse 214 marks the beginning of a revolution. He will how already described, sensed by magnetic head 197 and arrives at the line via controllable gate 212 215. From here, this impulse is sent to the controllable gate 203 via line 216 and to the controllable gate 218 and supplied via line 219 to the controllable gate 220. This will make these three Gates open. Immediately thereafter, pulse 221/1 from signal path 170 is sensed by magnetic head 190. One of the pulses 221/1 to is on this signal path at the end of each sector 186/1 to 186 / m 221 / m recorded, namely pulse 221/1 at the end of the sector 186 / m, pulse 221/2 on the other hand Recorded end of sector 186/1. The first sensed pulse 221/1 comes from magnetic head 190 via line 222 to the input of amplifier 223. After it has been amplified in this amplifier, it is fed to the controllable gate 202 via line 224, whereby this is also opened. Farther the pulse reaches the two controllable gates 227 and 228 via lines 225 and 226 The controllable gate 227 is also opened by this pulse, while the controllable gate 228 is used The pulse supplied via line 226 has no effect, since the starting position of this controllable gate so is that pulses are blocked. For the same reason, the pulse remains on line 229 of the controllable gate 228 has no effect, since the gate can only be blocked by pulses on this line. By opening the two controllable gates 202 and 203 by the pulses 221 and 214, it is now possible that pulses sensed by magnetic head 191 through amplifier 204 and these two controllable gates 202 and 203 get into the counting stage 201. At the same time, impulses get through one of the magnetic heads 193/1 to 193 / «can be sensed via amplifier 205 and the controllable gate 202 also in the counting stage 201. This is possible because one of the contacts 230/1 to 230 / m beforehand, d. H. closed before or when the contact 211 is closed by a control or similar device so that a connection between one of the magnetic heads 193/1 to 193 / w and the input of the Amplifier 205 consists. The pulses from magnetic head 191 and one of the magnetic heads 193/1 to 193 / m arrive shifted in time on line 210, so in each case one pulse from the corresponding one of the magnetic heads 193/1 to 193 / m is in a gap between two pulses from magnetic head 191. this will either achieved in that the magnetic heads 191 and 192 relative to the magnetic heads 193/1 to 193 / m be shifted by half a pulse width or that the recordings on the memory lanes 175 and 185 take place so that the pulses which are recorded there by half a pulse width compared to the recordings on the storage tracks 176/1 to 176 / «. Another The possibility is either in the transmission path from one of the magnetic heads 193/1 to 193 / m to the controllable gate 202 or in the transmission path from magnetic head 191 or 192 to this controllable gate stage 202 to incorporate a delay stage that adjusts itself according to the pulse position.

Since the sum of the pulses generated by magnetic head 191 and one of the magnetic heads 193/1 to 193 / m will be the sum of the last two digits of two numbers to be added Counting stage 201 is thus set to this sum. If the count of this partial result is greater than "9", then a pulse occurs on line 231, which occurs via the two open controllable gates 227 and 220 the control line 232 of the controllable gate 228 arrives. This opens it and it exists thus the possibility that a pulse applied to line 226 will pass this gate.

These pulses are entered in counting stage 201 while the first subsector 188/1 is passing. At the end of the passage of this subsector, after all of the recorded pulses of the memory tracks 175 or the corresponding memory tracks 176/1 to 176 / m have been counted in the counting stage 201, the magnetic head 189 senses the pulse 233/1 recorded on the memory track 178. This pulse is amplified by amplifier 234 and arrives at controllable gate stage 202 via line 235, which closes it, ie no more pulses can pass from amplifier 204 or 205 to counting stage 201. Furthermore, the pulse amplified by amplifier 234 is fed to controllable gate stage 227 via line 236 and to controllable gate stage 238 via line 237. This closes both gate steps, ie no impulses can pass. At the end of the passage of each subsector 188/1 to 188 / M ./. 1, the pulses 239/1 to 239 / n ./. 1 continue to be on signal path 181; draws. These pulses lag somewhat behind the pulses 233/1 to 233 / M ./. 1, ie after the functions just described have been triggered by the pulse 233/1, the magnetic head 195 of track 181 generates the pulse 239/1 on the subsector 188/1 sensed. This pulse is amplified by amplifier 240 and arrives at the controllable gate stage 242 via line 241. This opens it, ie pulses can pass through it. Furthermore, the pulse amplified by the amplifier 240 is fed via the diode 243 to the input of the counting stage 201, whereby this is counted further by one counting position, and at the same time this pulse reaches the input of the controllable gate 238 via diode 244 and line 245 was blocked by pulse 233/1, the pulse fed to line 245 cannot pass this gate. The controllable gates 238, 242 and 218 represent the transmission path for pulses which are transmitted by amplifier 246

amplified and recorded by one of the magnetic heads 199/1 to 199 / m . By opening the controllable gate 242 with the pulse 239/1, the possibility is created, after opening the controllable gate 238, to supply pulses to the amplifier 246 via these three mentioned controllable gates, since the controllable gate 218 has already been opened by the pulse 214. The controllable gate 238 is opened by a pulse which occurs on line 231 at the counting stage 201 and is fed to this controllable gate 238 via line 246. This pulse occurs when the counting stage 201 continues to count from the previously calculated result until it has reached its full counting capacity "10". This is done by pulses which are sensed by the magnetic head 194 from the signal path 180 while the first subsector 187/1 is passing and are amplified by the amplifier 247. These pulses pass via line 248 and the controllable gate 249 to the same input line 210 of the counter tube 201 are supplied to these pulses via line 250 to the input of the controllable gate 238th In the individual sub-sectors 187/1 to 187 / n ./. L, pulse groups 251/1 to 251 / m ./. 1 of "9" pulses each are recorded on signal path 180. These pulses are fed to the counting stage 201 in the manner described above and continue to count them, starting from the value set by the previous calculation and the pulse 239/1 , until they are zero. The possibility of supplying these pulses to the counting stage 201 via the controllable gate 249 is created in that pulse 239/1 is supplied to this via line 252 , whereby it is opened and thus allows pulses to pass. After completing the previous calculation, in which the last digit of two numbers to be calculated was added, pulses are now fed to the counter tube 201 until it has reached its full counting capacity. At this moment, a pulse occurs on line 231 , which is fed to controllable gate 238 via line 246 and to controllable gate 249 via line 253. This is blocked so that the further pulses amplified by the amplifier 247 are no longer fed to the counting stage 201 and the latter therefore remains at its zero position. The controllable gate 238, on the other hand, is opened by the pulse on line 231 , so that the pulses which exceed the maximum counting capacity of the counter tube 201 can now pass through the controllable gate 238 from amplifier 247 via line 250 and thus via the two controllable gates 242 and 218 can get to the record amplifier 246. One of the magnetic heads 199/1 to 199 / m is connected to the output of this amplifier via a closed one of the contacts 254/1 to 254 / m. The closing of one of the contacts 254/1 to 254 / m is carried out by a control together with the closing of one of the contacts 230/1 to 230 / w. By means of the corresponding one of the magnetic heads 199/1 to 199 / m, the pulses which are sensed by the magnetic head 194 and which exceed the maximum counting capacity of the counting stage 201 are recorded on the associated memory track 176/1 to 176 / m. Next, the pulse 255/1 is now sensed from the subsector 187/1 on signal path 184 by magnetic head 198 and amplified by amplifier 256.

This pulse is fed to the controllable gate 242 via line 257, whereby this is blocked for further pulses. This pulse is the signal for the termination of the first partial addition, during which the last digits of two numbers to be added are calculated. Immediately following the pulse 256/1 , pulse 221/2 is now sensed by the magnetic head 190 , which represents the signal for the start of the second partial addition of the penultimate digit of two digits to be added. This pulse arrives from the magnetic head 190 via the amplifier 223 - as already described - to the controllable gate 202 and further to the controllable gates 227 and 228. If a pulse occurred during the previous partial addition while the subsector 188/1 was being sensed on the output line 231 is, which indicates that the value stored in counting stage 201 is greater than "10", then this pulse reached the open gate 227 and the gate 220, which was also open at this time, as well as via line 232 to the controllable gate 228, which opened it. This open gate now offers the possibility for the pulse 221/2 amplified by the amplifier 223 to reach the input of the counting stage 201 via line 226, the controllable gate 228 and line 258. This is then counted out of its zero position at counting stage 1 and thus the tens carryover resulting from the previous partial addition is added to the following pulses recorded in subsector 188/2. In the manner described, the two numbers to be added are calculated one after the other, digit for digit. After almost one revolution of the drum 177, the sector 186 / M ./. 1 reaches the sensing position. The first digits of the two numbers to be added are recorded in this sector. They are also added in the manner described above. At the end of this sector, however, in subsector 187 / n ./. L, an additional pulse 259 is recorded on signal path 182 . This pulse is now sensed by magnetic head 196 and amplified by amplifier 260. It arrives at the controllable gate 212 via line 261 , which blocks the gate and thus prevents the start signal 214 from being sensed again. At the same time, this pulse reaches the controllable gates 218, 203 and 220 via the lines 262, 263 and 264, whereby these are brought into their starting position and blocked for further pulses. The following sector 186 / n was not used for computing or recording purposes, but control or other measures can be carried out during this time. If a subtraction is to be carried out with the device described above, switch 206 is switched from switch position “α” to switch position “b” . After contact 211 has been closed by a control, the calculation process described above takes place. By switching the contact 206 from switch position “0” to switch position “b” , the counting stage 201 is now fed via amplifier 204 not with the pulses 171/1 to 171 / M ./. 1 recorded in memory path 175, but rather with those recorded in signal path 185 Impulse groups 265/1 to 265 / m ./. 1. These pulse groups represent the complementary value to the count values for the number "9" recorded in memory path 175 in the corresponding sectors. The pulses are sensed by magnetic head 192 and fed to amplifier 204 via switch position "b" of contact 206. The rest of the calculation process takes place exactly as described above. This complementary recording makes it possible to use a subtraction as the addition of a count to the grain

ι uyt> ÜÖ4

perform the complementary value of a second counter value. This is done according to the following calculation scheme.

005738
-002593

+003145

005738
+997406

1003144

+003145

As can be seen, the subtraction of the number 2593 from the number 5738 gives the positive result 3145. In the case of the calculating device described above, this result is achieved by adding the number 997406 to the number 5738. This number represents the complementary value of the number 2593. A computing device with a six-digit input capacity was assumed as an example. The result of the addition 5738 + 997406 is 1003144. Since the capacity of the arithmetic unit only has six digits, the number 1 appears in the first position of the number 1003144 at the point in time after sector 186 / n ./. L has already passed the sensing position, ie at the time at which the pulse 259 has already been sensed by the magnetic head 196 and thus the actual arithmetic operation has been switched off. The aforementioned number “1”, however, generated a pulse on line 231 at the output of counting stage 201 during the preceding calculation process, which has opened gate stage 228 via the two open gate stages 227 and 220 . This makes it possible to feed the pulse 221 / w recorded immediately after pulse 259 , which was sensed by magnetic head 190 and amplified by amplifier 223 , via line 226 and the opened gate 228 to line 260 and from there to counting stage 201 . At the same time, this pulse arrives at the controllable gate 262 via line 261 , which opens the gate. This gate lies parallel to gate 212 closed by pulse 259 and thus enables pulse 214 to be sensed again, which starts a new calculation process. During this second computation process, the tracks 175 and 185 carry no records, since these were erased immediately after the first scanning by an erasing device (not shown). Thus, only the addition of the last pulse that occurred on line 260 with the result recorded on the corresponding one of tracks 176/1 to 176 / « takes place. The last sub-operation of the above calculation example is thus carried out and the actual result is saved on the corresponding one of the paths 176/1 to 176 / m. If the result of the subtraction is less than "0", ie negative, then the last pulse on line 260 no longer occurs and the calculation is ended after one revolution of the drum 177 , as the following example shows.

006743
-008271

-001528

006743
+991728

998471 ■

-001528

In this case, the result is a complementary value and is also saved as such on the corresponding path 176/1 to 176 / m. The conversion takes place during the writing process, which is not shown here.

With reference to FIGS. 12 and 13, the recording amplifier 246 and one of the controllable gates, e.g. B. the controllable gate 203 described. The other amplifiers used in FIG. 11 are basically of the same construction as that used in FIG. They differ only in the number of stages adapted to the individual conditions. The counting stage 201 is basically of the same structure as the counting stage 24 shown in FIG. 10. In the case of counting stage 201, only the two double triodes 127 and 128 are omitted. Instead of the counting stage described in FIG. 10, any other counting units can also be used in both devices according to FIGS. 1 and 11.

The controllable gate 203 shown in FIG. 12 contains the double triode 270 and the triode 271. The double triode 270 , together with its switching means, represents a flip-flop stage of known design, while the triode 271 represents a controllable amplifier.

This controllable amplifier is connected to the flip-flop in such a way that pulses which are fed to the grid 272 of the triode 271 on line 208 are either blocked by this triode or are amplified by it, and this happens depending on the position of the Flip-flops, ie depending on which system of the double triode 270 is live. The flip-flop is controlled by positive pulses which are fed alternately to the two lines 218 and 266. In the starting position of the flip-flop, the left system of the double triode 270 is live, ie a negative voltage drop occurs across the common series resistor 273 and also across the anode resistor 274. As a result, point 275 is strongly negative with respect to earth and thus also the grid 272, which is connected to point 275 via the grid bleeder resistor 276 , is strongly negative with respect to the cathode 277, since this is connected to earth potential. This negative bias on grid 272 is so great that positive pulses, which are fed to line 208 and reach grid 272 via capacitor 278 , cannot increase them beyond the response value of triode 271 , so that triode 271 is not modulated. If a positive pulse is now applied to line 216 , it arrives at grid 280 of triode 270 via capacitor 279 and increases the voltage on this grid so much that the flip-flop tips over and the left system of double triode 270 becomes live. The system on the right is blocked as a result, and at point 275 there is only one voltage difference with respect to earth, which corresponds to the voltage drop across the common series resistor 273 . This common series resistor 273 is dimensioned in such a way that the voltage drop occurring across it is just below the response value of the triode 271 . This makes it possible for positive pulses which reach grid 272 from line 206 via capacitor 278 to increase the bias voltage on this grid above the response value, so that an anode current increase in tube 271 occurs. This anode current generates negative voltage pulses at anode resistor 281 , which are fed to output line 210 via capacitor 282. As soon as a positive pulse is fed to line 263 , it arrives at the right grid 284 of the double triode 270 via capacitor 283 , whereby the flip-flop is switched back to its starting position and point 275 is again strongly negative with respect to ground.

As a result, triode 271 is blocked again.

13 shows the controllable recording amplifier 246. This amplifier contains the double triode 284, the left grid 285 of which is supplied from line 286 via capacitor 287 with pulses. These signals are distributed in the left system of the double triode 284

009 680/272

strengthens and fed to the right cutter 289 of the double triode 284 via capacitor 288.

After the signals in the right system of the double triode 284 have been amplified, they are passed through capacitor 289 fed to the grid 290 of the triode 291. The triode 291 can by the flip-flop, which the Double triode292 contains, can be controlled. The control happens in such a way that according to the Position of the flip-flop, the voltage on the control line 293 can be changed, d. i.e., the tension on this line 293 is either negative enough that the pulses arriving via capacitor 289 the grid voltage at grid 290 cannot rise above the threshold value of triode 291 and thus blocked, or else line 293 is less negative, so that positive pulses which are fed to the grid 290 via capacitor 289, the bias of this grid between the The response value and the value zero can change and are thus amplified by the triode 291. the Control effect of the flip-flop, the circuit and mode of operation of which can be assumed to be known, is based on the fact that across the common series resistor 294 and the anode resistors 295 and 296, respectively a voltage drop occurs, so that the anodes of the double triode 292 alternately less or less strongly negative to earth, depending on the switching position of the flip-flop. The initial position of the flip-flop is such that the left system of the double triode 292 is live. Then there is a voltage drop across the common series resistor 294 and also across the anode resistance 295. Both tensions add up, and at point 297 there is a strongly negative one Voltage drop which is transmitted to the grid 290 of the triode 291 via the grid bleeder resistor 298 will. If the right grid of the double triode 292 is now a positive on line 299 via capacitor 300 Pulse supplied, then this is switched to its starting position, and now the right System energized while the left system is locked. As a result, there is one less at point 297 strongly negative bias, since the voltage drop across anode resistor 295 is now only through the cross current of the voltage divider, which is formed from the resistors 301 and 302, caused will. Since the cross current of this voltage divider is very low, there is a small voltage drop via anode resistor 295, and at point 297 there is essentially only the negative voltage, which is generated by the common series resistor 294. In this case, the impulses which are fed to the grid 290, amplified by the triode 291. Should the flip-flop be in its starting position switched back and triode291 blocked again, then the left grid becomes the Double triode 292 is supplied with a positive pulse from line 303 via capacitor 304, whereby the right system is blocked and the left system becomes conductive again. The pulses amplified in triode 291 are fed to the grid 306 of the triode 307 via capacitor 305. This grid is about resistance 308 negatively biased against cathode 309. The anode of the triode 307 is connected to the anode resistance 310 and one of the magnetic heads 199/1 to 199 / m on earth potential, while cathode 309 with -150VoIt is connected. About the corresponding of the magnetic heads 199/1 to 199 / m continues to flow the cathode current of the triode 311, since the cathode 312 This tube is also connected to earth via the connected magnetic heads 199/1 to 199 /? i. That Grid 313 of triode 311 is connected to the anode of triode 307 via grid bleeder resistor 314. Furthermore, the grid 313 is via the resistor 315 at point 316, which is the connection point of the Voltage divider, which is formed from the two resistors 317 and 318, shown. The lower connector of resistor 318 is connected to control line 293. This enables the grid 313 in Depending on the voltage of the control line 293 to give a different bias. Prevails a strongly negative voltage on control line 293, which corresponds to the blocking of triode 291, then becomes the voltage at grid 313 via resistor 315 is also so negative that the anode current through Tube 311 is blocked, i.e. H. becomes zero. On the other hand, if the control line293 becomes a less strongly negative one If a bias is applied, then the grid 313 is also less strongly negative with respect to the cathode 312, see above that a relatively strong anode current can flow. This anode current flows through the switched on of the magnetic heads 199/1 to 199 / w and generates a constant magnetic field through them. This serves to any existing recordings on the memory path associated with the corresponding magnetic head Clear. A new recording of pulses by this magnetic head 199 takes place in that when driving the tube 307 by positive pulses which are supplied to the grid 306, the anode current rises through this tube and thus a negative voltage drop occurs across anode resistor 310. This negative voltage drop is caused by the grid bleeder resistor 314 on the grid 313 of the triode 311 transmitted so that the anode current through this tube is greatly reduced or blocked. Now Only the anode current of the triode 307 flows through the magnetic head 199. Since this current is reversed The direction in which the previously existing anode current of the triode 311 flows, this corresponds to a reversal the polarity of the magnetic field generated by the magnetic head 199 and thus a non-magnetization of the Magnetic layer currently in the drawing position. This differently directed magnetization of a small part of the storage area corresponds to the recording of a pulse.

The principle of offsetting binary numbers will now be shown with reference to FIG. The basic principles of the invention also remain in the offsetting of binary numbers, namely in places Offsetting of two numbers, the staggered supply of the impulses that represent the digits, and receive the sequential accounting and recording of the result. Fig. 14 includes the Drum storage 320, of which part of the outer surface is shown. This carries the two storage tracks 321 and 322 as well as the two signal paths 323 and 324. In this device, too, the scope is of the drum store divided into individual sectors 325/1 to 325 / m. Each of these sectors 325/1 to 325 / m is in turn in four subsectors 326/1 to 326 / m, 327/1 to 327, 'n, 328/1 to 328 / «and 329/1 to 329 / m divided. Each of the subsectors 326/1 to 326 / re is dimensioned so that it has a storage capacity of one bit contains. In the sub-sectors 326/1 to 326 / m and 329/1 to 329 / m are on the signal tracks 323 and 324 sync pulses recorded. In sub-sectors 327/1 to 327 / m and 328/1 to 328 / m, on the other hand, the information to be calculated is recorded on the memory lanes 321 and 322. Storage path 321 is in subsectors 328/1

up to 328 / w the first digit to be added in binary representation recorded. This can also be a result of a previous calculation operation. In the sub-sectors 327/1 to 327 / w, on the other hand, have the second summand recorded on memory path 322. Thus, the impulses are the individual digits in each digit of the two numbers to be offset represent, offset from each other by a certain pulse interval. The records on the Memory tracks 321 and 322 are sensed by the two magnetic heads 330 and 331 and shared The amplifier 333 is supplied via line 332. From there you can get to the via control line 334 controllable gate 335. This controllable gate has the task of generating impulses which are sent to subsectors 329/1 to 329 / "are recorded on signal path 323 and sensed by magnetic head 336 and by amplifiers 337 are reinforced, either to let through or to block. The pulses are received from amplifier 337's Input line 338 of the controllable gate 335 is supplied. The circuit of this gate is such that the pulses, which are fed to control line 334, this gate can either open or close, correspond the previously prevailing switch position. The starting position of the controllable gate 335 is as follows that pulses applied to line 338 cannot get to output line 339, i. H., the gate is closed. After the first pulse, which is sensed by the magnetic heads 330 and 331, respectively and was amplified by amplifier 333, the control line 334 was fed, the controllable switches Gate 335 so that a pulse, which is fed to input line 338, is sent to equalization line 339 can be removed, d. that is, the gate is opened. By the second one fed to control line 334 On the other hand, the controllable gate 335 is closed again. On control line 340 you can Pulses are supplied from the controllable gate 341 which always open the controllable gate 335, i. H. always switch over so that pulses from line 338 can reach line 339. The two controllable Gates 341 and 342 are designed to be either closed or opened by a pulse depending on which of the control lines 343 and 344 or 345 and 346 these pulses are fed. Impulses which are fed to the control lines 344 and 346, respectively, cause an opening, and impulses which are fed to the control lines 345 and 343, cause a closing of the corresponding gate 341 or 342. The controllable gate 341 is used to generate pulses which recorded on the subsectors 326/1 to 326 / «on the signal path 324 and by magnetic head 347 are sensed and amplified by amplifier 348, either to block or the control line 340 feed. The controllable gate 342 serves the purpose of impulses which occur on line 349, either to block or to feed the control line 344. A pulse then occurs on line 349 when the controllable gate 335 is switched from its open position to the closed position. Because the starting position of the controllable gate 335 is in the closed position, then always occurs on line 349 Pulse on if more than one pulse was fed to the controllable gate 335 on control line 334, d. i.e., the pulse on line 349 represents a carry pulse. The recording of the calculation result is done by magnetic head 350 which receives pulses from line 339 via record amplifier 351 are fed. Record amplifier 351 is based on the principle of the record amplifier shown in FIG built up. Amplifiers 333, 337 and 348 represent normal amplifiers such as the shown in Fig. 3.

The mode of operation of the device will now be described using the following calculation example:

1 / υ \ 1 0 Γ. 10 1 1 1 0 1 1 1 10 1 1 1 + - 10 1 11

1 0 1

During one revolution of the drum store 320, the magnetic head 347 first generates the synchronization pulse 352/1 sensed. This arrives via amplifier 348 on control line 346 of the controllable Gate 342, which opens it. Since the starting position of the controllable gate 341 is that pulses are blocked, this pulse cannot reach the control line 340 of the controllable gate 335. At the same time, this pulse arrives from amplifier 348 via control line 346 to the controllable one Gate 342 and opens it. Then, by the magnetic heads 330 and 331, the two records 353/1 and 354/1 sensed. These reach control line 334 via amplifier 333. Through the Recording 353/1 generated pulse becomes the controllable Gate 335 opened and closed again by the one generated by recording 354/1. At the Closing the controllable gate 335 occurs on line 349 an impulse, which over the previously opened controllable gate 342 arrives at control line 344 of controllable gate 341. This will open it. In the subsector 329/1, the recording 355/1 is now sensed by the magnetic head 336 and amplified by amplifier 337. Since the controllable gate 335 is closed again by recording 354/1 this pulse cannot get from amplifier 337 to line 339. Rather, it just gets over Control line 345 is fed to the controllable gate 342, whereby the gate is closed. Since the controllable Gate 335 was closed, no recording was made during this partial addition in sector 325/1 by magnetic head 350, as is also expressed in the calculation example given above. The function sequence just described corresponds to the addition of the last digit of the two to be added Counting. The result of this partial addition is: 1 + 1 = 10. This makes the result zero characterized in that the recording 354/1 is erased by magnetic head 350, since the controllable gate 335 is closed, however, no new recording will be made by magnetic head 350 as previously described instead of. The result recorded in sector 325/1 on memory web 321 is therefore zero. the "One" of the result occurs as a carry pulse on line 349. The possibility of consideration at the next partial addition, the impulse appearing on line 349 is created the controllable gate 341 opens, so that a pulse is now fed to the controllable gate 335 via line 340 which switches this to position »1«. Position »1« corresponds to the position in which it is possible to feed pulses from line 338 to line 339. The reverse Switching position, namely that the controllable gate 335 pulses which are fed from line 338 ^ locks, corresponds to the "0" position. The feed

a pulse on line 340 is effected by sensing signal 352/2 at the beginning of the sector 325/2. The pulse generated by this signal in magnetic head 347 passes through amplifier 348 and the controllable gate 341 to line 340 and thus brings the controllable gate 335 to position "1". Simultaneously this pulse is fed via line 346 to the controllable gate 342, whereby this is opened to to allow a carry pulse that may occur during the subsequent partial addition to pass. Subsequent to the record 352/2, the magnetic heads 330 and 331 become the record 353/2 and 354/2 sensed. These reach the input line 334 of the controllable via amplifier 333 Step 335, whereby this is initially switched back to position "0" by pulse 353/1 and thereby a carry pulse is generated on line 349. As previously described, this opens the controllable one Gate 341. The pulse generated by the recording 354/2 puts the controllable gate 335 back into its Position "1" so that now the pulse generated by recording 355/2 in magnetic head 336 and is amplified by amplifier 337, from line 338 to line 339 and thence across Recording amplifier 351 can reach the recording head 350. The recording head 350 is with respect to the sensing head 330 by the width of one of the subsectors 326 to 329 in the direction of rotation of the drum store 320 shifted. This means that when through the magnetic heads 330, 331, 336 and

347 the subsectors 329 are sensed, recording head 350 opposite the subsector 328 is located on storage lane 321. The pulse sensed by magnetic head 336 of subsector 329/2 355/2 accordingly generates a recording via recording head 350 in sector 328/2. This corresponds to the state "1" of the memory and thus the result in the second digit of the result of the above given calculation example. The pulse appearing on line 338 reaches the control line at the same time 345 of the controllable gate 342, whereby this is closed, so that a pulse which is on line 349 arises, this gate cannot pass. This pulse on line 349 arises from the fact that the pulse occurring on line 339 via diode 355 of control line 334 of controllable gate 355 and this switches back from its switch position »1« to switch position »0«. Through this a pulse occurs on line 349, but this is not a real carry pulse generated by the calculation represents and therefore must not have any influence on the position of the controllable gate 341. The blocking of this pulse by the controllable gate 342 is achieved in that the controllable Gate 335 tilts a little slower than the controllable gate 342, so that it is already closed until the line 349 a pulse occurs. Record 352/3 is now sensed at the beginning of sector 325/3 and the generated pulse via amplifier

348 supplied to the controllable gate 341. Since this is done by the first on line 349 of the sector 325/2 occurring impulse has been opened, the impulse from amplifier 349 can this controllable gate 341 pass and arrives at the controllable gate 335 via line 340, which brings it to position "1" will. This corresponds to the carryover resulting from the previous partial addition. Simultaneously gate 341 is closed via control line 343. The controllable via amplifier 348 The pulse supplied to gate 341 also reaches the controllable gate 342 via line 346, whereby this gate is opened and thus offers the possibility of a possibly arising during the next partial addition To supply the carry pulse to the controllable gate 341. This next partial addition in sector 325/3 consists of the fact that the record 353/3 to the. Carry stored in controllable gate 335 is added, d. i.e., the record 353/3 is sensed by magnetic head 331, in amplifier 333 amplified and fed to the controllable gate 335 via line 334. This will put it in its zero position switched back and gives a pulse via line 349 to the open controllable gate 342 Control line 344 of the controllable gate 341. This is opened. Now by magnetic head 336 sensed the recording 355/3 and the pulse generated thereby via line 338 the controllable Gate 335 supplied. Since this is in its switch position »0«, the impulse is blocked. Simultaneously This pulse reaches the controllable gate 342 via line 345 and closes it. At the beginning of Sector 325/4 is now recorded 352/4 and sensed via amplifier 348 as a pulse controllable gate 341 supplied. Since this is open, this pulse reaches line 340 and thus brings the controllable gate 335 in switch position "1", which is the result of the previous partial addition resulting carry. The controllable gate 341 is closed again via line 343. the Subsequent partial addition of the fourth from last digits of the two numbers to be added consists of the addition "0" + "0" + "l". The "1" results from the previous carry. Thus, after passing the Sub-sectors 327/4 and 328/4 put the controllable gate 335 in switch position "1", and that by recording 355/4 Pulse generated in magnetic head 336 comes from line 338 through line 339 and record amplifier 341 to recording head 350, where it reads the "1" corresponding to the result on memory path 321 in subsector 328/4 records. At the same time, this pulse, which is sent to the control line via diode 356 334 arrives, the controllable gate 335 is switched back to its "0" position. The resulting As already described, an impulse on line 349 cannot pass controllable gate 342. Since the The controllable gate 341 remains closed. The now following pulse, which is generated by the recording 352/5, thus arrives only via control line 346 to controllable gate 342 and opens it. In the subsequent partial addition of the first digits of the two summands, the two recordings 353/4 and 354/3 are sensed, and the controllable gate 335 is initially set to switch position »1« by the pulses generated thereby and then switched back to switch position »0«. This again produces a on line 349 Transmission pulse which is fed to the controllable gate 341 via the open gate 342 and this also opens. The pulse subsequently generated by trace 355/5 on line 338 will be locked by the controllable gate 335, so that the recording in this sector is "0", as well as off can be seen in the calculation example. Then, in a sector no longer shown, 325 / w + l the subsequent recording 352/6 is sensed, which via line 348 sends a pulse to the controllable gate 341 and via this to line 340 delivers. This moves the controllable gate 335 into its switching position »1« toggled. Since in the storage lanes 321 and

According to the calculation example in the now following sector 325 / «+ 1 there is no more recording, the controllable gate 335 remains in its switch position» 1 «and a pulse, which is generated by the subsequent recording 355/6 , passes through it to the recording amplifier 351. A recording is thus effected in sector 325 / w + 1 on the storage path 321 by the magnetic head 350. The result is, as already shown in the calculation example, »10101 0«.

As can be seen from the example, the computing capacity of such a device is only determined by the capacity of the drum store 320 . Of course, any other, cyclically operating memory type can also be used instead of drum accumulators. Instead of the purely binary operating device shown here, decimal-binary operating devices can also be built according to the invention.

Claims (8)

Patent claims: 1. Electrical computing device in which the pulse trains representing the decimal places of a number are supplied in series to form sums counters and in which the result is recorded on a rotating magnetic memory, characterized in that a pulse train with such a time delay compared to the other pulse train is fed to a counter that originally simultaneous pulses arrive at the counter one after the other, and that the head recording the result is offset by a recording length of the decimal number from the sensing head of the path of one summand, so that the result is used at the same recording location as the one Summand was taken from the invoice. 2. Computing device according to claim 1, characterized by an additional electrical delay device, which is supplied with a series of pulses which is derived directly from the recorded pulses and represents the numerical value will. 3. Computing device according to claims 1 and 2, characterized in that the pulse trains a Clock track can be taken. 4. Computing device according to claim 3, characterized in that to decrease the offset Pulse trains from the clock track offset from one another by half a pulse interval Heads are provided. 5. Computing device according to claims 1 to 4, characterized in that gate circuits (22, 23) are connected to the heads for removing the pulse trains, which lead to the counter (24) and which are opened by initial pulses and closed by end pulses. 6. Computing device according to claim 5, characterized in that the initial pulse by the End of a number of recurring signal recordings or through the beginning of a such signal series or is determined by the fact that a pulse has its amplitude, direction, frequency or phase changes. 7. Computing device according to claims 1 to 6, characterized in that the pulse trains from Pulses are derived from the area assigned to the relevant places by the number of the magnetic memory are recorded, and that the recorded pulses in represent the numerical value to be processed for each area part assigned to a point. 8. Computing device according to claims 1 to 7, characterized in that the pulse trains are recorded in several equivalent tracks (176 / 1-176 / 1) and that a transition is made from one track to the other when executing the calculation. Considered publications: German patent specifications No. 966 461, in particular pp. 9 and 12, 936 066; British Patent No. 731 140, particularly page 3; U.S. Patent No. 2702666, particularly page 1; "Arithmetic Operations in Digital Computers", D. van Nostrand Comp., Inc., New York, 1955, p. 339 to 341, p. 237; "The Annals of the Computation Laboratory of Harvard University", Vol. XXV, Cambridge, 1952, Pp. 10-11; Vol. XXVII, Cambridge, 1951, p. 184. In addition 4 sheets of drawings © 009 680/272 12.60