US2484081A - Electrical item comparing system - Google Patents

Description

Oct. 11, 1949. A; H. DICKINSON ELECTRICAL ITEM COMPARING SYSTEM 9 Sheets-Sheet 1 Filed April 1'7, 1945 Z INZENTQH BY Z W ATTORNEY Oct. 11, 1949. A. H. DICKINSON ELECTRICAL ITEM COMPARING SYSTEM 9 Sheets-Sheet 2 INVENTQR ATTORNEY .wmmm nw u m n .7 PW v m n N u w.. a m P...

III

Filed April 17, 1943 1/ In I Oct. 11, 1949.

A. H. DICKINSON ELECTRICAL ITEM COMPARING SYSTEM 9 Shets-Sheet 4 Filed April 17, 1943 z 2 ryzumn.

ATTORNEY.

Oct. 11, 1949. A. H. DICKINSON EL|ECTRICAL ITEM COMPARING SYSTEM 9 Shee'ts-Sheet 5 FilPd April 17, 1943 jmdE Y a z VENTOR. B

ATTORNEY.

Oct. 11, 1949. A. H. DICKINSON ELECTRICAL ITEM COMPARING SYSTEM 9 Sheets-Sheet 6 Filed April 17, 1943 mxmm whmm

I z T R.

ATTORNEY.

Oct. 11, 1949.

A. H. DICKINSON ELECTRICAL ITEM COMPARING SYSTEM 9 Sheets-Sheet '7 Filed April 17, 1943 WNN Nmm Emww IN VEN TOR.

A TTORNE Y.

wmw mww SQ QQ vm Q&

Q Q H A 2 Oct. 11, 1949.

A. H. DICKINSON ELECTRICAL ITEM COMPARING SYSTEM Filed April 17, 1943 I 9 Sheets-Sheet 8 @nmw Emww

INVENTOR ALI'TORNEY Oct. 11, 1949. A. H. DICKINSON 2,484,031

ELECTRICAL ITEM COMPARING SYSTEM Filed April 17, 1943 M I 9 Sheets-Sheet 9 D98765432I0H 9 76543210 98765 a Et- Eu- Ex- D981654-32l0I/D98765432I lID A'TToRNEY Patented Oct. 11, 1949 ELECTRICAL ITEM COMPARING SYSTEM Arthur H. Dickinson, Scarsdale, N. Y., assignor to International Business Machines Corporation, New York, N. Y., a corporation of New York Application April 17, 1943, Serial No. 483,506

19 Claims. 1

This invention relates to an electrical system for determining the relative magnitude of values represented by electrical signals. The values may be digits or alphabetic characters or other symbols to which value significance may be applied. The values may be represented on suitable value sources such as settable value registers or record material on which the values may be represented by a suitable code.

Means heretofore utilized for comparing values and determinin their relative magnitude have utilized moving parts whose mass necessarily produced inertia and momentum effects which limited the operating speed Of these devices. Further, because of the existence of such mass; noise, wear, and other undesirable efiects were produced which could be reduced by careful design but not eliminated.

The general object of the present invention is to provide such comparing means as will be practically inertialess and will embody solely electronic means having negligible inertia which may be disregarded completely in practice.

Another object is to provide electronic machine control means controlled by electronic comparison of values.

Another object is to provide electronic means for electronically, selectively manifesting comparison conditions.

Another object is to provide electronic means for comparing multicolumnar data.

Another object is to provide a plurality of electronic comparing means so related that a manifestation by one will supersede a manifestation by another of :the comparing means.

Another object is to provide electronic means capable of comparing alphabetic data, as well as numerical data.

Other objects of the invention will be pointed out in the following description and claims and illustrated in the accompanying drawings, which disclose, by way of example, the principle of the invention and the best mode, which has been contemplated, of applying that principle.

In the drawings:

Fig. 1 is a cross-sectional view of the record controlled mechanism.

Fig. 2 is a sectional view taken along line 2-4 of ma Fig. 3 is an enlarged sectional view taken along line 3-3 of Fig. 1.

Fig. 4 is a wiring diagram of one of the electronic trigger circuits employed in the invention.

Fig. 5 is a wiring diagram of another electronic trigger circuit utilized in the invention.

Figs. 6a and 6b comprise the wiring diagram of the record controlled embodiment of the invention.

Fig. '7 is a timing chart for the latter embodiment.

Figs. 8a and 8b comprise the wiring diagram of a modification of the invention.

Fig. 9 is a timing diagram for the modification.

The invention is disclosed in connection with a main embodiment and a modification. The main embodiment is a record-controlled machine and includes means to compare data designated on records and to control the machine accordin ly. The modification comprises means to com pare data entered in or represented by data representing means. Such data representing means may beconsidered, for illustration, as manually settaole, but it will be evident that automatically settable data representing means may be used.

In both forms of the invention, electronic means are provided to respond to data from corresponding columns of two data sources and to control an electronic comparison manifesting device for the columnar order. The columnar comparison manifesting devices control a common comparison manifesting device according to comparison conditions in all the columns; that is, the common manifesting means manifests the comparison relation of the data in all orders.

In the record-controlled embodiment, the common comparison manifesting means will control continuation of operation of the machine or may control another desired operation automatically and selectively according to a comparison result. Further, a plurality of such control devices are embodied in the record-controlled machine, each relating to a different class of data. These 6011-. trol devices are so related that one will dominate another.

In the modification, the control device will be disclosed as manifesting by visible signals the difierent comparison conditions. It will be evident that other suitable manifestations may be effected.

3 The record-controlled embodiment will now be described.

1. General-record controlled machine Before proceeding to a detailed description of the completely novel contro1 device, a more general description is set forth. Each of the plurality of control devices comprises a number of electronic tubes and related circuits for comparing data and for manifesting the agreement or non-agreement of data. The manifesting means comprises discrete electronic units, also termed elements, each element including electronic tubes. A number of such elements, termed denominational elements, equal to the number of columns of data to be compared are provided. A common element, termed a class element, is provided, for effecting an all-column manifestation under control of the elements which relate tothe separate columns. Each element may have either an on or an off status. When an element is in off status following comparing operation, it indicates a condition of data agreement. When an element is in on status following a comparing operation, it indicates a condition of data disagreement, The status of the denominational elements is regulated under control of record controlled entry means producing differentially timed impulses, each indicative of a determined entry.

It is appreciated, in view of the foregoing, that, if two multicolumnar items are compared and agree, all elements are in oif status, and that, if the items disagree, one or more elements are in on status. The class or common element be comes turned on, when one or more of the denominational elements are in on status. The common element, therefore, when off, manifests a condition of agreement between the two items of data, and when on, manifests a condition of disagreement. The status of the common element controls the entry means whereby the latter is maintained in operation as long as the common element remains in off status and is interrupted in its operation when the common element is in on status.

In man instancesit is desirable to effect control of a record controlled entry means by more than one class of classifying data and generally similar data. For this reason, therefore, a control device is provided in connection with each class of data. Each device is capable of manifesting the agreement or non-agreement of its related class of data and is also capable of maintaining or interrupting the operation of the entry means.

Since classifying data and generally similar data may be divided into main and subordinate classes, provision is made in this invention whereby the manifestation of the control device related to the main class of control data supersedes the manifestations of the control devices related to the subordinate classes of control data and Whereby the manifestation of the control device related to the first subordinate class of control data supersedes the manifestation of the control device related to the second class of control data.

The record medium for controlling the record controlled entry means comprises a film upon which multidenominational control or classifying data and other data are recorded by means of code designations such as spots located at different positions. The spot representations are scanned by light to, control photoelectric cells in electrical circuits whereby differentially. timed electrical impulses are produced, indicative of the control data to be entered into the control devices.

The film, upon which the differential spots are located is controlled by a feeding mechanism whose operation is manually initiated but is automatically terminated. This mechanism includes two analyzing stations separated b a distance which is slightly less than the length of one film frame. From both of the first and second analyzing stations, classifying data are entered into the control devices. In this manner the items comprising classifying data upon two adjacent film frames are compared simultaneously for agreement or non-agreement. As long as all classifying data on successive film frames agree, film been chosen.

a freely mounted on the sleeve.

feeding continues. When there is a disagreement between either the main or any of the subordinate classes of control data on adjacent film frames, film feeding is interrupted so that the type of control break, main or subordinate, may be observed. Film feeding is then resumed after one manual operation. This conditions the extra elements of the control devices which manifest a break in classifying data and initiates film feeding operations, which automatically continue until another disagreement in the classifying data on adjacent film frames is manifested.

2. Structurerecord controlled machine (Figs. 1 to 3) Thi structure includes means to feed records and analyze them for data. The record medium is a film F of which successive frames constitute successive records. Fig. 2 shows five such frames designated l, 2, 3, 4, and 5. The feed means is driven, through clutch means, by a motor M mounted on a base 50. An upright frame membar 5! is carried by the base and supports the feeding and analyzing means. All of the structure is enclosed by a casing 52 having a door 53.

The film is wound on a supply reel 54 and fed by three similar pairs of feed sprockets 55 to a take-up reel 56. The film is suitably guided and formed with loops as shown in Fig. 1. A single clutch may be used to drive all three sprocket pairs. However, to reduce stress and inertia, a separate clutch is. used for each sprocket pair. This clutch may be of any suitable construction and one such as disclosed in Patent 2,150,227 has Since the details of the clutch do notv enter into the invention, it will be briefly explained only in sofar as necessary to understand its present function. It may be mentioned that while the clutch is used in said patent as an accumulator actuator, its function here simply is to releasably connect a sprocket pair to the motor Mv for drive. Briefly, the drive part of the clutch is a rigid assembly of gear and ratchet 6| journaled on a rod 62. Both the gear and ratchet have ten teeth. The driven part of the clutch includes a sleeve 63 j ournaled on rod 62 and to which one sprocket pair is fixed. A detent disk 64 is fast onv the sleeve while a toothed disk 65 is Carried, in a manner not shown here, by disk 64 is a clutch dog 66. A clutch lever 61 engages a tooth of disk 65 to prevent rotation of the driven part of the clutch. When the clutch lever is rocked clockwise (Fig. l) by energization of a clutch magnet 68, it releases disk 65. The disk thereupon permits clutch dog 66 to engage ratchet 6| which then drives the sleeve 63 and-the parts carried engagement with disk 65 and, as the disk 64 with clutch dog 66 move slightly further, the clutch dog is forced to disengage ratchet 6|. The driven part of the clutch is thereby uncoupled from the drive part. Gears 60 of the clutches are meshed with gears I0. The lowest gear I is driven by gearing II from motor M. One-to-one gearing, including a gear I2 (see Fig. 1), transmits drive from the lowest gear ID to the other two gears 10. Rigid with gear 72 is a pulley I3 connected by a coil spring belt I4 to a pulley I5 on shaft I6 of the take-up reel 56. The reel is thereby urged constantly in take-up direction.

Feed sprockets 55 are of such size that in one tenth of a revolution they feed the film through one frame length. Since feed sprockets 55 are clutch-driven by ratchets BI, it follows that onetenth of a revolution of the ratchets 6| effects one frame advance of the film. As stated before, each ratchet 6i has ten teeth, so that one-tenth of a revolution thereof is its travel through one tooth distance. Thus, one tooth travel of the ratchet will cause a single frame advance of the film. This single frame advance is taken here as the measure of a machine cycle (see Fig. '7). When the clutch dog 66 is permitted to engage the ratchet 5 I, a tooth of the ratchet will pick up the clutch dog at the beginning of such machine cycle, and in one machine cycle, the ratchet will move through one tooth distance and cause advance of the film for one frame length.

As the film feeds from the upper pair of sprockets 55 to the next pair, it moves through a film gate. The film gate comprises a channeled guide plate I1 and a coacting pressure plate I8. The plate I! is fixed to one end of a block I9, while the plate 78 is slidably carried by a confronting end of a block 80. Both blocks are of molded material and attached to frame member 5|. The gate may be opened by retracting the pressure plate I8. This is efiected by manually rocking a lever 8| clockwise. The lever is connected to the pressure plate and is acted on by a spring 82 which normally maintains the pressure plate in closed position. In its passage through the film gate, the film is scanned for data designations. Such designations are in the form of spots disposed in differential positions of a film frame and represent numerical data by the single point Hollerith code and alphabetic data by a known combinational code. The analyzing means includes upper and lower designation sensing means. These are spaced apart slightly less than a frame length. Each sensing means is of the light responsive type including photocells and a light source. The cells of the upper sensing means are designated UP and those of the lower sensing means, LP. Upper and lower groups of quartz rods 84 and 85 are molded in the block 8!]. The upper group pertains to the upper analyzing means and the lower group to the lower analyzing means. Each rod terminates at its front end (right end) in alinement with a column of the film. The rods 84 thus terminate in a single transverse line at one side of the upper sensing station, while rods 85 are similarly alined at their front ends at one side of the lower sensing station. Each rod 84 conducts light passing through a designation position in one column of the film, at the upper sensing station, to one cell UP. Similarly, each rod 85 conducts light passing through a designation position in a column of the film, at the lower sensing station, to one cell LP. The quartz rods 84 and 85 constitute one side of the upper and lower sensing means, respectively.

The other, coacting side of the sensing means includes a common light source consisting of an elongated lamp 86 carried by a socket 81 which is attached to frame member 5|. Molded into the block I9 are two wide quartz pieces 88 and 89. Piece 88 conducts light from the lamp to the right side of the upper sensing station While piece 89 conducts light to the lower sensing station. Each piece forms a thin line of light at its related sensing station. This line of light at a. sensing station is such as to cross all the columns of the film and to illuminate only a single row of the designation positions when the row is at the sensing station. In this manner, at each sensing station, light is commonly supplied to all the designation positions of a single row, while light passing through a position in any particular column is conducted by one quartz rod to the related photocell. When a designation position in a column bears a data representing spot, light to the related column cell is obscured. In a manner described later, the reaction of the cell to the change in light manifests the significance of the spot.

It will be clear that while one record or film frame is being sensed by the upper scanning means, the preceding record or frame is being scanned by the lower sensing means. Further, each frame will be scanned successively by the upper and lower sensing means.

It will be recalled that in one cycle, the film is advanced through one record or frame length. This cycle has fourteen cycle points (see Fig. 7). Thus, a record will pass a sensing station in one cycle of fourteen cycle points. Further, the designation positions of a column of a record will pass a sensing station in successive portions of the cycle, each such cycle portion being equal to the cyclic spacing between successive cycle points.

3. Oscillatorsrec0rd controlled machine Pulses are utilized in this invention for various purposes. One pulse may be required in any of the fourteen cycle portions which are defined by the cycle points of a cycle. Accordingly, pulses must be supplied at the same rate as the frequency of the cycle points. The source of these pulses is preferably a main oscillator of the type known as a multivibrator. The duration of time corresponding to one cycle determines the base frequency of this oscillator, as will be clear from the previous explanation. Essentially, the multivibrator consists of a two-stage, resistance coupled amplifier in which the output of the second stage is fed back to the input of the first stage. Such an oscillator is capable of producing either square-topped or saw-toothed waves, depending upon the portion of the oscillator from which the waves are derived. The square-topped waves are employed herein because they are easily changed into pulses of extremely sharp wave front and short duration. The circuit diagram of the multivibrator and its principle of operation will now be described in detail.

Referring to Fig. 6a, a voltage is supplied to lines I30 and I3I, and to a voltage divider consisting of resistances I32, I33, I34, and I35. Potential is also supplied by means of this divider to lines I36, I31, and I38, their potentials being positive with respect to each other in the order given and with respect to line I3I, The oscillator, generally designated W, comprises a duplex tube containing triodes I39a and I39b and associated resistances and condensers. The an- 7 odes of the respective triodes are connected to line its through load resistors a and Mill) and the common cathode is directly connected to line I3I.

The anode of triode I3 9a is coupled back to the grid of triode i39b by means of coupling condenser Mib which is also connected to line I3! through the grid leak resistance I422). The anode of triode I392) is coupled back to the grid of triode I lIa, which is also connected to line I3I through the grid leak resistance MZa. With this circuit connection, the normal bias of the grids of triodes I39a and I39b is zero. Such an arrangement is unstable and oscillations are initiated by I a minute change of emission of either triode. Assuming that the current through I39a momentarily increases, there is produced across resist ance Mta an increased voltage drop and a decrease in potential across [3961. The resulting negative pulse is fed through coupling condenser I lIb to the grid of triode I392), making it more negative. Current through I391) is decreased, decreasing the voltage drop across resistance l mb and an increase in potential across I39b. This increase is equal to the original decrease across 939a multiplied by this tubes amplification factor and is thus much higher. Coupling condenser Mia conveys this potential change to the grid of 539a making said grid much less negative with a resulting rapid increase in the current through I39a. The voltage drop due to this increased current is in turn fed to I391) with cumulative re sults. Actually, the current fiow through I390: is

increased to a high valuesubstantially instantay;-

neously, which flow reaches a maximum when the grid of H391) has a negative potential great enough to reduce the current flow through. 139?) to. zero. When this condition is reached, the

charge on condenser Mlb commences to leak off It will now be understood that a heavy current iiows alternately and for a given period of time through each of the triodes I39a and I39b. When triode E3911 is conducting, the other triode, i39b, is shut off; this situation then instantaneously reverses and I3a is shut off and tube i391) conducts. This produces alternate and sustained voltage drops across resistors IMla and hiiib, these voltages being 180 degrees out of phase with each other. These voltages are in the form of squaretopped waves, easily converted into pulses that possess a steep wave-front and are extremely short in duration.

Fig. '7, line a, diagrammatically illustrates the voltage oscillations (with respect to line IZ-SI) which occur at point MBa (Fig. 6a) and shows that they are square-topped in form and occur fourteen times per machine cycle. Fig. '7, line b illustrates the voltage oscillations (with respect to line ifiii) which. occur at point H531) (Fig. 6a) and also shows that they are square-topped in form and also occur fourteen times per cycle. Since as stated above these voltages are 180 degrees out of phase the potential of point 643a rises at each of the fourteen cycle points and drops midway between cycle points while the po- I39a by means of coupling condenser tentlal of point I-43b rises midway between cycle points and falls at each cycle point. One cycle of oscillator operation is that period between successive potential rises of point M311, for example, and its time duration in seconds is propor tional to the sum of the time constants of condenser I4Ia and resistor MM and of condenser Mill) and resistor I421), respectively, i. e., 1/ is proportional to R14zaC141a+R142tC141b. It is understood that while the frequency is determined primarily by the grid-leak resistance and gridcondenser capacity, it is also influenced by the remaining circuit constants, the tube characteristics and the electrode voltages.

A rise in potential of point I i3b causes charging of condenser I45 and current flow through resistor I45 to line I38. By suitably choosing the values of resistor I 45 and of condenser I so that their R. C. product is relatively small, the rise in potential of I43?) produces on I45 a positive pulse of extremely short duration having a steep wave front. A decrease in the potential of M32) causes I44 to discharge and a negative pulse of the character just noted is thereby produced on M5. Since the rise and fall of I432) is constantly recurring, positive and negative pulses are continually produced on resistor I45, of the form shown in Fig. 7, line 0.

The foregoing has described the manner in which an oscillator of the multivibrator type is employed to produce square-topped waves, which are converted into pulses of extremely sharp character, suitable for use in various portions of the circuit of the control device. The manner in which pulses are employed for control purposes is described subsequently in Section 5. The synchronization of an auxiliary oscillator by means of pulses generated by the main oscillator is now considered.

Attention is directed to the fact that whenever film feeding operations occur, a film frame completely traverses a sensing station in one machine cycle. It is sufficient to state here that each differential position on the film must be at a sensing station at the time a control pulse is generated. Accordingly, the film feeding mechanism, when it is operating, must be synchronized in operation with regard to the generation of pulses by the main oscillator. That is to say, during one machine cycle in which one film frame traverses a sensing station, it is required that the rate of presentation of the differentially disposed designation positions to the sensing station be equal to the rate at which the main oscillator generates control pulses.

It has already been pointed out in Section 2 that a movement equal to one tooth movement of ratchet BI (Figs. 2 and 3) advances the film one frame length. It is therefore apparent that the time during which ratchet BI is advanced one tooth must be equal to that of a machine cycle. Accordingly, the motor M (Fig. 1) is driven at the proper speed to cause ratchet SE to advance one tooth each machine cycle. Synchronism between the film feeding mechanism and the generation of control pulses by the main oscillator is accomplished by supplying motor M (Figs. 1 and 6a) from a power source which is controlled from said main oscillator.

It has been explained above that positive and negative pulses appear on resistor I45. The grid of triode I 462) is connected to resistor I 45. The cathode of I461) is connected to line I36 and resistor I45 terminates at line I38. The difference in potential between lines I36 and I38 is the grid bias for triode I46b and is sufficient to maintain the triode at shut-off when no pulses appear on resistor I45. It is therefore apparent that negative pulses appearing on I45 momentarily serve to increase further the grid bias of 14% without affecting the shut-off condition. A positive pulse, however, on resistor I45 reduces the grid bias of triode I461), permitting current to flow therethrough momentarily. A resonant circuit, comprising inductance I41 and condenser I48, in series with triode I461), between lines I30 and I36, is momentarily excited upon current flow in the triode, and oscillations are initiated in the resonant circuit. The inductance I41 and condenser I48 are adjusted so that the period of oscillation is either equal to, or is a multiple or submultiple of the frequency with which I461) passes current. The output of this resonant circuit is coupled by means of condenser I49 and resistor I50 to an amplifier II. The output of this amplifier is of sinusoidal character and of suflicient power to drive motor M, when switch I52 is closed. In this manner, a power supply of correct frequency is ailorded the motor so that when the film feed means is subsequently clutched to the motor drive, the feed of film is such that the rate of presentation of differentially disposed designation positions to a sensing station equals the rate of generation of control pulses by the main oscillator.

Having described the main oscillator which produces control and synchronizing pulses and the auxiliary oscillator for supplying power to the motor of the film feeding mechanism. a fundamental circuit employed in the control device is now discussed.

4. Trigger circuit A basic circuit, which is employed in this invention comprises electronic discharge means, illustrated in Fig. 4 as a plurality of triodes in one envelope, these triodes being associated with resistances and condensers as shown. These triodes are so interconnected in a circuit and operate in such a manner that the circuit assumes two conditions of stability. When one of the triodes is conducting, a large amount of current flows throu h it and the other triode is at shutoff. In other words. in one condition of stability, one of the triodes has a relatively low impedance and the other has a relatively high impedance. In the other condition of stability, the respective conditions of the two triodes is reversed. Controlling impulses are applied to resistances and condensers associated with the circuit to cause the shift from one condition of stability to the other. Every other impulse brings the circuit to the ori inal condition of stability. Such an arrangement of vacuum tubes and associated circuits is herein termed a trig er circuit, and voltage variations therein, which are determined by the conditions of stability, may be employed for various controlling purposes.

Referring to Fig. 4, voltage of the polarity in dicated is supplied to lines I30, I3I and I36 in the manner described in Section 3 in connection with Fig. 6c.

The trigger circuit (Fig. 4) comprises two impedance networks. One network includes resistances 1611a. 1611:. and I621), resistance I6Ia being shunted by coupling condenser I63a. Triode I641) is connected in parallel between the junction of resistances I6Ila and 16!!) and line I36. The second impedance network consists of resistances I661), I 6Ib and I621), resistance I6Ib being shunted by coupling condenser I631). The triode I640) is connected in parallel between the junction of resistances Ib and I 611) and I 36. Resistances 166a and I601) are equal in value as are resistances I Glu and I6I1), and resistances 62a and I621). The capacities of condensers IBM and I631) are also equal. In actual practice an efficient combination was found when the values of resistances IBM and IBM were each approximately one-third the value of i6Ia. A suitable value for the capacity of I631) is of the order of a few hundred micromicrofarads.

Assuming that the grid of I641) is substantially at the potential of cathode line I 36, its grid bias will be substantially zero. With resistance I601) properly chosen, 164a has an impedance relatively low as compared to that of I661), and its anode and point !651) to which the anode is connected will have a voltage which is not much greater than that of cathode line I36 with large current flow through I641). With resistances I6 I1) I 621) proper-1y chosen, the potential drop across 16Ib is great enough to maintain point I66?) and hence the grid of tube I641), negative with respect to cathode line I36. With I641) negatively biased, it has an impedance greater than that of 166a. Hence the anode of I641) and point 165:) to which the anode is connected are at a high enough potential so that the voltage drop across resistance I6Ia will not force the potential of point I66a below that of line I36. The foregoing describes one condition of stability of the circuit in which I64a has a large current flow therethrough and I641) is at shut-off; hence, with no current flow therethrough, and point I65a is at a higher potential, with respect to lines I36 and I3I, than is point I651). The manner of switching the trigger circuit to the other condition of stability, is as follows.

A Condenser I61a is connected in series with a resistance 168a and the latter terminates at point Ifita. A condenser I611) is connected in series with a resistance I681) and the latter terminates at point I661). Condensers 161a and I611) serve to couple resistors I681). and I681) respectively to sources of steep pulses, not shown in Fig. 4. In the absence of any pulses on resistors I660. and I681), the potentials of points H561; and I661) remain at the values determined by the condition of stability described above.

When a positive pulse is applied through condenser I611) to resistance I681), it is also applied to resistor I621), since these two resistors are connected in series. Accordingly. point I561) rises in potential with respect to line I 3!. Also, the diiference in potential between cathode line I36 and I 661). the grid bias of triode I641) is reduced. If the pulse applied to 1661) and 6 1) is of sufiicient value, the grid bias voltage of I641) is reduced to less than the shut-off value. S ch operation permits I641) to pass current. causing point 55a. to suddenly drop in potential. producing a negative pulse. This pulse is fed through condenser !63a to the grid of IBM, effecting a sudden increase in the negative grid bias thereof, and reducing current flow through I64a and resistance 1613b. Point I651), accordingly, rises in potential, with respect to line I 36 to produce a positive pulse which is fed through condenser I 6 31) to the grid of I641), changing its grid bias to substantially zero. Since, as just described, the potential of I651) has now risen and that of 165a has now dropped, triodes I64a and I641) assume a condition of stability which is the reverse of that originally described; namely, I64a is now at shut-offwh-ile I641) passes a large amount of current. The new status of the trigger circuit is maintained until a positive pulse is ap plied to resistances I631! and Itiia. When this occurs, the resulting negative grid bias reduction of l-ii ia causes increased current flow therethrough, and the trigger circuit is returned to the former condition of stability.

In the foregoing description, it wa assumedthat positive pulses are not applied concurrently to resistances I881) and H521), and to resistors H386! and I62a. If, on the other hand, either positive pulses from two sources are concurrently applied through condensers Ifila and Hill; to associated resistors or positive pulses from a single source are applied with I-tla and IEi'Ib connected in parallel as indicated by the dash line wiring IGiL-such pulses are effective alternately only, in two branches of the circuit and cause it to shift back and forth from one condition of stability to another.

Assume now that pulses are simultaneously applied to two points of the trigger circuit by either of the two methods given in the paragraph above and that the trigger circuit is in the con-- dition' of stability where points 55a and I552) are'respectively high and low in potential. When a positive pulse is applied through condenser Iii'la toresistances I681; and time, point Ififia rises somewhat inpotential with respect to line I35. Since H5611 is already substantially at the potential of line I36, the bias of triode 564a is at or near zeroand it is passing substantially maxi-- mum current. Therefore, a further rise of Ifitain potential may cause the bias of I641; to go slightly positive without affecting current flow thereth-roughl- Thus a positive pulse: applied to this branch of the trigger circuit, when it is in the condition of stability specified, has no effect thereon. The simultaneous application of a positive pulse through condenser I6-Ib to resistors I68!) and I621), however, does cause the trigger circuit to shift its condition of stability in the manner described above. In further explanation, the positive pulse applied through condenser I'SIa and resistor I'Bta to the grid of triode IBM is extremely sharp whereas the negative pulse applied through condenser I B-Sa is exponential in character'. The positive pulse will, therefore, have substantially lost its eiiect when the negative pulse applied via con-denser I631]. is still effective. It requires only a minute change in conductivity of triode H541) to start the triggering action which continues in a regenerative manner until the circult is fully triggered in the same way as when a positive pulse is applied only'to the grid of triode I6-4b. Just as the circuit is triggered, when positive pulses are applied concurrently, from the status in which point I'S'Eib is at low potential to the status which this point is at high potential, the circuit may be triggered back to its former status when the positive pulses are next applied concurrently to the grids of the triodes.

So far in this description of the trigger circuit. positive pulses have been employed to cause a shift in the condition of the circuits stability. It is to be noted that negative pulses are equally effective in causing a shift in the condition of stability. Assume that the condition of the trigger circuit is such that points IE?) and I65a are respectively high and low in potential. When a negative pul e is applied through condenser It???) to resistor I681), it is also applied to resistor I621). Accordingly, point I651) falls in potential. with respect to line I3 I, and the grid bias of triode I641;

iii

12 is increased. Such operation reduces current flow through triode I641), causing point I-65a to suddenly rise in potential, producing a positive pulse. This pulse is fed through condenser I63a to the grid of IBM, effecting a sudden decrease in the grid biasthereof, and causing increased current flow through Ma and resistance I601). Point ISEb, accordingly, drops in potential, with respect to line I36, to produce. a negative pulse which is fed through condenser I632) to the grid of I641], increasing. its grid bias to. shut-off value. Since, as just. described, the potential of I652), has now dropped and thatv oi I65a has now risen, 564a and Lfi Ib assume a conjoint condition of stability which is the reverse of that originally described,

namely, I64a passes a large amount of current While I641) is shut ofil. The new status of the trigger circuit ismaintained until another negative pulse. is. applied to I68a and I62a.. When this occurs. the resulting. grid bias increase of IBM causes decreased. current flow therethrough, and the trigger circuit is returned to first condition of stability.

In the description just concluded, it was assumed that negative pulses are not applied con.- currently to resistances IBBb and 5%, and the resistors I581; and. IBM. If, on the other hand, either negative pulses from two sources are concurrently applied through. condensers 161a and I Bib to associated resistors or negative pulses from a single source are applied to IBM and I-B'lb, connected in parallel as indicatedv by the dash line wiring I69, such pulses are effective alternately only, in two branches of the circuit and cause it to shift back and forth from one condition of stability to another.

Assume now that negative pulses are simultaneously applied to two points of the trigger cirsuit by either of the two methods given in the previous paragraph and thatv the trigger circuit is in the condition of stability where points Hi5-h and Ififia are respectively high and low in potential. When a negative pulse is applied through condenser IB'Ia to resistances IBM and I62a, point Ibfia .drops somewhat in potential with respect to line I.3I. Since Nita is at the lower of its two possible values, the bias of tube ISM is equal to or greater than that required for shut-off, when IMa passes no current. Therefore, a further drop in potential. of iSBa merely causes an increase in the grid bias. of I540. without afiecting the shut off condition. Thus a negative pulse applied to this branchv of the trigger circuit, When it is in the condition of stability specified, has no efiect thereon. The similar application of a negative pulse through condenser ISIb to resistors I58?) and "32b, however, does cause the trigger circuit to shift its condition of stability in the manner described above. In further explanation, the negative pulse applied directly to the grid of triode I644! is extremely sharp whereas the positive pulse applied through condenser I63a is exponential in character. Accordingly, the negative pulse will have been substantially dissipated when said positive pulse is still eflective. The triggering action thus continues and is completed in the same way as though a negative pulse were applied directly only to the grid of triode I642).

That portion of the circuit in Fig. 4, within the broken line enclosure is a trigger circuit which is used in various portions of both the main embodiment and the modification. For purposes of simplification, this enclosed portion also may be termed an element which may be considered to be on when points I651) and I 65a are at high and low potentials, respectively, and to be off when the potentials of I65b and I65a are low and high, respectively. The voltages which exist at points, such as I65a and I651), of the triggering circuit, and which vary in accordance with the conditions of stability are employed for many control purposes as subsequently explained. Several such elements are utilized in the invention. To aid in identifying the purpose of parts of these elements, these parts will be designated the same as in Fig. 4. Pulses are applied to some of the elements via condensers and resistors such as IB'Ia and I6") and H581; and I681) of Fig. 4. Ready identification of the purpose of these parts will be afforded by designating the similar parts wherever they appear by the general reference numbers I61 and IE8 followed by one or more letters a or b.

It may be noted that to successfully achieve the operations as described above, the pulses applied to the points ISBa and IE6!) of an element should be of steep wave form. Preferably the R. C. product of the value of resistances such as IBM and I620. and the value of the capacity of an associated condenser such as IBM should not exceed one-fifth the R. C. product of resistance IBIa and condenser I63a. The effect of these relative time constants is to prevent a single pulse from causing more than one change on the state of the element or trigger circuit.

The foregoing has described an element comprising a trigger circuit in which vacuum tubes and electrical parts are so interconnected and operated as to produce alternate conditions of stability. Either positive or negative pulses may be employed, and these may be derived from either a common source or a plurality of sources and applied to two points of the circuit. Pulses simultaneously applied are effective alternately only in two branches of the circuit and cause it to shift back and forth from one condition of stability to another.

5. Control device-record controlled machine A general explanation of the operation of the control device as applied to a record controlled machine is given in Section 1 and, therefore, need not be repeated here. Accordingly, the specific details of the construction and operation of the control device are now set forth.

Prior to placing the machine in operation the operator closes switch (not shown) which supplies a voltage to lines I30 and I3! (Fig. 6a) and also to lamp I92. Certain conditioning operations are next performed and these are described in detail subsequently (Section 6). For pur-- poses of facilitating the description it is assumed that the conditioning operations have been performed, as a result of which elements '1, At, An,

Am, Bt (Fig. 6b), Bu, B03, Ct, Cu and Car, each of the same nature as the trigger circuit element described in Section l, are in off status. Accordingly, the point I651) in each of the aforesaid elements is at low potential with respect to l3I. Since the main and intermediate oscillators (Sec tion 3) now function, the operator closes switch I52, placing motor M in operation.

Referring to Figs. 1 and 2, the film F is inserted in the record feeding mechanism (Section 2) so that the leading edge of frame I is about to pass the lower sensing station. With such adjustment, it is seen that frame 2 is about to pass the upper sensing station I I0.

Referring now to Fig. 6a, a voltage divider comprising resistors I89 and I90 is connected between lines I39 and I3I. Normally, switches I92, I93, I94, and I are in the position shown, thus affording circuit connections for condensers I96, ti l, 598, and I99 between a point 299 on the aforementioned voltage divided and I3I. circuit arrangement and with the switches positioned as shown, the condensers I96 to I99 are each charged to a potential equal to that across resistor I90.

To initiate film feeding operations, the operator throws switch I93 to reverse position from that shown. The discharge of condenser I91 through resistance I621) causes point I961) of element T to rise in potential, thus shifting element T to an on status. This rise of potential of point I661) occurs, by happenstance, shortly after 9 of cycle I (Fig) 7, line d). The grid of a triode M611 is connected to point I66b of element T, the anode of the triode is connected to line I30 by parallel clutch magnets 68 (also see Figs. 1 and 2), and. the cathode of this triode is connected to line I39. The rise in potential of point I591) of element T reduces the grid bias of triode Idiia causing current flow therethrough and through the clutch magnets 68. The energization of these clutch magnets releases the several clutch dogs 66 for engagement with ratchets 6| (Fig. 3) of the clutch mechanism (Section 2). Since rotation of the ratchets through one tooth distance requires a complete cycle, dogs 69 will not be picked up by the ratchets until D following the switching of element T to an on status and regardless of the time at which, by happenstance, this element is switched on. Since element T is a trigger circuit such as described in Section 4, when it is switched on its point I651) remains at relatively high po-- tential and triode I461; remains conductive and clutch magnets 68 remain continuously energized, until element T is switched to off status.

It wil1 be assumed that automatic control is to be effected according to comparison of major, minor, and intermediate classification data recorded in six columns of the film.

Fig. 6a shows the six cells LP and six cells UP to sense the classification data at the lower and upper sensing stations. Each cell is similarly connected into a control circuit which determines the current flow through an electronic unit. The units relating to the two cells LP and UP which sense corresponding columns are in parallel with each other and connected to a common load resistor. Impulses are produced on this resistor upon changes in current flow through the connected tubes. These impulses control the switching of a deinoninational control element from one status to another. There may be a plurality of denominational control elements for each class of data. Thus, if the major class has two columns of orders, there will be two denominational control elements for such class. The denominational control elements relating to the same class each control a common class control element. The condition of the class element determines whether an agreement has been found in all the orders of the class of data. Thus, there will be one class control element for the major class, one for the intermediate class, and one for the minor class. The major class control element dominates the intermediate and minor class control elements, and the intermediate element also dominates the minor class element.

As stated in Section 2, the lower and upper sensing stations are slightly less than a frame length apart. It follows that with respect to two like designation spots on successive frames, the spot With this on the leading frame will be sensed slightly before the spot on the following frame. In other words, since one cycle is required for travel of each frame across a sensing station, the spot on the leading frame will be sensed at a time of the cycle Which is just previous to the time at which the spot on the following frame will be sensed. With the film adjusted in the manner previously described in this section, a spot on the leading frame will be sensed substantially at an exact cycle point while a corresponding spot on the following frame will be sensed a slight fraction of a cycle portion later.

For convenience, in the further description, parts may be distinguished as follows: The letters U and L denote relation of the parts to upper and lower sensing means; letters A, B, C denote major, intermediate, and minor class relations; and letters t and it denote tens and units orders.

In Figs 6a and 6b, the major, intermediate, and minor class control elements are designated, respectively, Ar, Br, and Cr. The denominational control elements associated with element Act are designated At and Au. Similarly, the denominational elements associated with the intermediate class element 1350 are designated Bt and Bu, and those" associated with element C2: are designated Ct and Cu. The control triodes for the denominational. element At are designated LlAt and UlAt. The common load resistor connected to these triodes is designated EAL The similar intermediate class parts are designated LlBt, UlBt, and Hit, and the similar minor class parts are designated LlCt, UiCt, and 2Ct.

The manner in which each photocell shown in Fig. 50. controls a related triode is the same and will be explained in detail only in connection with the cell LPt (major) which senses the tens order of the major field of the leading film frame.

The cell is connected by a plugwire 884 to a resistance I85. This resistance is tapped at a chosen point 186 by a wire it! leading to the grid of triode LIAt. The cathode of the triode is connected to line I36 while resistance I85 terminates at line [31. The cell is of the type whose impedance is lowered by light falling thereon. When the cell is illuminated, its impedance is low and the potential at point we is substantially the same as that of line The grid bias of triode LlAt is then approximately zero, and there is substantially maximum current flow through the triode. When light to the cell is obscured, the cell impedance is high, and the potential at point I86 is approximately the same as that of line ltl. The grid bias of triode L-lAt is then high, and current flow in triode is rapidly curtailed. It should be understood that the cell reacts extremely rapidly to a change light falling thereon and causes a correspondingly rapid change in the cur-- rent flow through the related tube and, hence, of its impedance.

As the impedance of triode LlAt rapidly rises, the potential across the load resistor 2At sharply drops. The load resistor is connected to condensers I 61a and lii'ib which lead through resistors lBEa and its? to points ififia and I661) of denominational control element At. It should be noted that condensers iiila and 51b and resistors 58a and 4 correspond to the similarly designated parts of the circuit shown in Fig. 4 and explained in Section 4. Hence, upon the occurrence of the drop in voltage across the load resistance 2At, the condensers apply steep positive pulses to both branches of element At, shifting it from assumed off state to an on state. The above explains the action of a control element in response to the sensing of a designation at the lower sensing station. Subsequent to this action, a matching designation may be sensed at the upper station. In a manner now clear, the related triode UiAt will increase rapidly in impedance, and a second further sharp drop in voltage across the load. resistance 2At will occur. Positive pulses are again produced by condensers iiila and E632; and element At thereupon shifts back to 01f state.

Attention is directed to the fact that because of the triodes LlAt and UiAt being in parallel with each other, the voltage across the load resistor rat will depend on the impedance of both triodes. Thus, when one triode rises in impedance while the other triode remains at low impedance, a first drop in voltage across the load resistor occurs. When the impedance of the other triode also rises, a second, further drop of voltage across the resistor occurs. Likewise, a decrease in impedance of one triode, while the other triode remains at high impedance, causes the potential across the resistor to increase to a,

first extent, and a decrease in impedance of both triodes produces a further rise in potential across the resistor.

It has been explained that when a designation on the leading frame was sensed, element At was turn-ed on and that the element Was turned 01f upon the sensing of a matching designation on the following frame. The operation of turning element to one status and then back to the original status may be referred to as a switching sequence. Thus, the sensing of the matching points jointly controls one switching sequence (elf to on and back to off) of a related denominational control element. A second, similar switching sequence of the element, under joint control of the matching designations as they successively leave the lower and upper sensing stations, is eiiected as follows. When the spot in the leading frame leaves the lower station, light again falls on the related cell LP and the grid bias of triode LlAt decreases rapidly, causing the impedance of the triode to drop abruptly. The result of this is a first rapid rise in potential across the load resistor, causing negative pulses to be applied by condensers iiila and ifi'lb to element At. In the manner described in Section 4, such pulses also shift the status of the element. Since the element. at the end of the first switching sequence, was off, it is now shifted to on status again. Subsequently, the spot on the following frame leaves the upper sensing station and the cell UP is again illuminated. The impedance of the related tube UlAt drops rapidly, and there is a second, further rise in potential across resistor fiAt. Consequently, negative pulses are again applied to element At, returning it to off status. This compietes the second switching sequence of the denominational control element under joint control of matching spots.

It may be mentioned at this time that lines 11, u, 5 k, n, q, r, u, y, and z of Fig. 7 graphically represent high and low potential states of points i661) or points i551) or on and ofi states of the trigger circuits.- or elements, the on state being indicated by the high part and the off state by the low part of the graph relating to an element. Lines e, f, i, and other lines of Fig. '7 relating to eFectronic discharge units, such at LlAt, graphically represent high and low current.

The operations will be explained further with reference to the illustrative data tabulatedbelow. The general (accounting) data are included merely to: indicate other data-whichmay be recorded on the film.

While the 9 spot in thetens order of themajor. field or". frame l-..is-tr.aversing. the lower station ,..it interrupts lightto. the. related. cell LB; sharply. reducing current .fiow. through? the. triode .InAt.v This action occurs at the 9.1"timeof the cycle II, assindicatedby line-e (ii/Fig; 7 Theincreased: impedance of. triode IiiAtu causes substantially,

immediate switching of. element Atfromzofl to $011.

state (see line 9. otFig. 7,.cyole II) ,inthe manner explained before.

Since frame. 2.:contains-a: matchingQ spotin the. tens column .ofzi. the; major. .field, the. light to the related, cell vis interrupted and the current flow through triodeUlAtreduced.rapidly.. This,v action occursuafterv the: element At: hasshifted; to on state, as jindicatedsbyi comparison'oflines andg. of Fig..7,.oycle IL, The increasedimpedance of triode UlAtcauses a. second shift of; theeIement At, so that-the element returns to off state. This completes one switchingsequence of the element- Referenoe toline g of Fig. 7, cycle II, shows that element At, after completion of the first switching sequence, remains in off status mid-way of the cycle. portion between cycle points 9 and 8. Suchattained status :is a manifestation of matching data in corresponding columns of two successive film frames. The configuration of a. designation spot is such that in its passage through a sensing-statiomit interrupts the light to the related photocell for slightly more than half the cycle portion between? two successive cycle'points. Thus, the9'spot in the tens order of the major field of frame I leaves the lower station shortly after the. midway point between the cycle points sandfi. The related cell LP is again illuminated at. this time andsthe impedance of triode LlAt-ra-pidly drops, causing the voltage drop=acros-s resistor 2At-to increase rapidly. As a result; condensers lfi'laand 1 61b apply steep negative pulses .toelement At, shifting it again toonstate at. the time indicated in line g of Fig. 7, cycle. II.

With regard to frame 2, the matching 9-spot leaves the upper sensingistation shortly before- 8, permitting light to vfall again on related cell UP. A sharpdrop in; impedance of triode'U-lAt" results, andthe voltage drop across resistor ZAt rapidly increases to a further extent. negative pulses are again applied topoints-Ififia and I662: of element At, shifting it back to off status. This completes -the second switching se quenceof element At .and the element remains in oil? status until designations-on:frames-2 and 3 are compared. It is-seen that following a point Steep midway between cycle points, a second I complete 1 switching sequence, occurs when there is an"- lower and uppersensm stations 3 at the mid-index p'oin't' time.

18* cQI'unir isof successive frames. A subsequent swit ng 'ofthe element to on and then to ofi status} prior-to the -end of the cycle portion, ocours as the matching spots successively leave the The net resultis that the el'e'rnentis oil at the end of the cycle portion and is ready for operation in accordancewvith'the rieiit comparison of data in the related columns; When thedesignations on the frame-sbelng' eomparec d'b'riot' agree, these designations will be sensed in different cycle portions. Thefir-stdesignation-sensed will be the one corresponding to a higher value; Each designation, when-sensed, willturn on' the denominational elenient w'hich will then stay on until this same designation leaves the sensing station. The desigii'aticm spot is of such length as to require more thari'h'alf a c ycle portion to cross the sensm station. Hence, the element will remain on In a manner explained later, this-condition of the element at such; mid-indexpoint time causes a disagreement of data to be manifested." This manifestation will :be eff'ectedasrthe higher value representing? spot of the two non-matching spots crosses-asensing station. Such spot may be either on the leading-or the" following frame. When? theispot leaves the stationafter the midindex point time the' element is again turned oif: Thus, a switching: sequence will be completed under'individual control of each spot dur ing each of the-separate cycle portions in which such non-matching spots are scanned. Each such switching sequence requires a longer time thanthe-switching sequ'ence elfected under control of"matching:designations. Before the first of the individually controlled switching sequences is completed'the element is' on at a mid-point and a condition "of non-agreement will be manitested. 7 The subsequentcompletion of the switchingsequenoebrings the'element to off state, in conditiontotoperate under control of the next spoton spots: in the relatedwcolumns of the successive frames The manner in which the on state of a denominational'cont-rol elementcauses a condition of non-agreement; i. e;,-a"break in classification-data; .to-be manifested will be explained shortlyvhereafter.

Reference to the above-table of illustrative classifying data indicates" that the remaining classifyingdataon frames l and 2 agree; i. e., 87654'"-and "87654-.-. Inview of the detailed explanation givenregarding'operation of element At under'a condition of-agreement in designation-spots. a detailed'description of the operation ofthe other control elements under conditions of agreement is believed, unnecessary. It is sufiicientto statethat the twoswitching sequences are effected for each of the other control elements at the times indicated by lines 9', n, q, u, andy. It is clear now that all the control elements At; An, Bt; Bu, Ct, and'Cu are in off status at midpoints. of indicated-cycle portions, manifesting-agreement of major,- intermediate, and. minor data inframes-l and- 2.

Point lfili'b'of element Atfl is connected to the gridof a tube -3At Whenelement At is off, point 5551s at low potential and the grid bias of the tube3Atis high. Under this condition, there is substantially. no current flow through the tube; In a simila'r'manner, points i665 of the other control elementsarefconnectedto the grids of related tubes. Aswill :be' xplained later in this section;Iwlin'currerit"flowdo'es'exist in one or more of these tubes at mid-points of cycle portions, conditions of non-agreement in classifying data are manifested and one or more of the elements A02, Br, and C: are shifted to on status. It is suflicient to state here that since all classifying data on frames I and 2 agree, elements A03, B23, and Cr remain in off state during the machine cycle in which these frames are compared. When these elements thus remain in ofi state, feed of the film continues and a third cycle will be performed.

During the third cycle, frames 2 and 3 will be compared. Since frame 2 contains a 9 spot in the tens order of the major field, element At is first shifted to on state, such shift occurring at the 9 point of cycle III, as shown by line g of Fig. '7. Frame 3, however, contains a 4- instead of a 9- in the tens order (major field) and there is no interruption of the light falling on its related photocell UP (Fig. 6d) at shortly after 9. Accordingly, element At remains in on status at midway between 9 and 8 of cycle 3, as indicated by Fig. 7, line g (III). With ele-'- ment At on at this time, its point I661) is at high potential, and therefore, a low bias voltage is applied to the grid of its associated tube 3At, causing current flow through said tube and a common load resistor 4A. The screen grid (hereinafter termed screen) of a pentode A is connected to a point 206 of a resistor 4A. The current flow through tube 3At and resistor 3A brings about an increase in the potential drop across the resistor and, accordingly, the screen voltage of tube 5A is reduced.

The control grid (hereinafter termed grid) of tube 5A is connected to line I36, as is the cathode of this tube. With this circuit arrangement, tube 5A is always maintained at zero bias, and variations in current flow therethrough are controlled solely by changes in its screen voltage. Since this screen voltage is now low, current flow through tube 5A and its load resistance 6A is relatively small and, therefore, the drop in potential across the latter resistance is small. The screen of a tube IA is connected to point 269 of resistor 6A. The grid of tube IA is connected to wire 2 I 0 which extends to a point on resistance I45 on which, as explained before in Section 3, is continually produced the steep pulses, as shown in line 0 of Fig. 7. The cathode of tube IA is connected to line I35 and resistance I45 terminates at line I38. The difference in potential between lines I36 and I38 is the normal grid bias of tube 1A. This grid bias is sufiicient to maintain this tube at shutoff for the values of screen potential which are utilized. It is only when both the screen voltage of tube IA is at high value and the grid bias of said tube is reduced by a positive pulse appearing on resistance I45 that there is current flow through the tube.

When the drop in potential across resistance 6A is small, as is now the case, the point 209 and, hence, the screen voltage of tube IA, is at a high potential. The appearance of a positive pulse on resistance I45 at midway between 9 and 8 (Fig. '7, line 0 III) reduces the grid bias of tube "IA (Fig. 6a) causing current flow therethrough. The anode of this tube is connected to point I65a of resistance IGIJa of element Am. The current flow through tube IA is in the form of a sharp pulse, and causes an extremely rapid increase in potential across resistance I600. of element Ax. Accordingly, a negative pulse is applied by the condenser I63a of the element to the grid of triode I64a of the element. The ourrent'flow in triode 'I64a falls rapidly, point I65b rises in potential, and the grid bias of triode I64b is reduced sharply. Consequently, current flow in triode I641) rises, maintaining point I65a at low potential. Since point I651) is now at high potential and point I65a at low potential, the element Am has been switched to on status. The time at which this action occurs under the stated disagreement condition is indicated in line It, cycle III of Fig. '7. As has been mentioned previously (Section 1), the manifestation of the control device related to the major class of data supersedes the manifestation of the devices related to the intermediate and minor classes of data. Also, the manifestation of the intermediate control device supersedes that of the minor control device. The means whereby such superseding control is effected is now set forth.

It has been explained above that element Arr: has been shifted to on state as a result of a disagreement found in the tens order of major class data on frames 2 and 3. As is now understood, point I651) of element A1: is high in potential when the element is turned on. Point I651) of element A0: is coupled by pairs of condensers I6'Ib1) (Fig. 6a) and resistors I68bb and wires 366 and 30I (also see Fig. 6b) to points I66b of elements Ba: and Car, respectively. Point I651) of element Ba: (Fig. 6b) is also coupled to point I661) of Cat: by means of a condenser I6lbb and a resistance I68bb and the wire 30L The recovery time of the condenser-resistor circuit is relatively small and, therefore, the rise in potential of point I651) of element Ax at midway between 9 and 8 of cycle 3 causes steep positive pulses to be applied concurrently to points I661) (Fig. 6b) of elements Ba: and C02, thus shifting both of them simultaneously to on status at the time indicated in lines 1* and z of Fig. '1 (III).

To recapitulate, when frames 2 and 3 were compared, a disagreement Was found in the tens orders of the major classification data designated on these frames. This condition was manifested by the retention of an on status by element At (Fig. 60.) during a. mid-point between two cycle points. With element At in on state at such time, the result was that element As: was turned on. As element Aa: was switched on, it brought elements B1: (Fig. 6b) and Ca: to a like status. Accordingly, whenever there is a manifestation of a break in the major classifyin data, such manifestation produces break manifestations for the intermediate and minor classes of data, even though there should be no disagreement in either of the latter classes of data. It will be appreciated that, by virtue of the point I651) of element Ba: and point I661) of element Ca: being coupled by a condenser I6'Ibb and resistor I68bb, when there is a break in the intermediate classifying data, such manifestation produces a like manifestation for the minor class of data. The break in the intermediate data, however, has no effect on the manifestation of the major control device. Further, when there is a break solely in the minor classifying data, th manifestations of the major and intermediate control devices are unaffected.

As the -9 spot, recorded in the tens order (major field) of frame 2, leaves the lower sensing station shortly after midway between 9 and 8 of cycle 3, light again falls on its related photocell LPt (Fig. 6a) and the current flow in triode LlAt is increased. Hence, element At, previously in on state because of the discussed disagreement condition, shifts to an 01f status in a manner now under-stood. The resulting; increase in current.

flow through triode LlAt. and the shift of ele ment At. to an oil status. occun slightly after mid.-. way between? and 8, as. indicated by lines e and 91, Fig. 7, cycle Ill. Thus, only a single switching sequence. is; effected. during one cycle portion (between successive cycle points). when there is a disagreement of, data in corresponding columns of successive film frames. It is. clear that a denominational controlielement. is switched to an on status and remains in said status midway of. acycle portion to indicate a. conditionoi noneagreement, and" is. automaticall witched off prior to the. termination of the cycle portion, under control of the non-.matohed data spot of one only. of the successive film framesbeing compared. By. thusautomatically.returning. said con.-

rol. elementto ofi. state after. it has. manifested av condition of non-agreement. of data, theelement is. placed. in proper status for operationdn follow L ing. cycle portions.

The operations resulting from a; condition. of non-matching datain theitens orders of-the major data of frames! and Zhave beendescrihee. It has beenshown thatwhen the datav in corresponding columns. of. two successive film frames dis agree, a denominational. control element At uh dersoes. one complete switching operation. only, thereby. resulting in its; being in. on. status at a id point. of aeyole. portion. Further, it has been sho n at. it is the. onistatus of. the denomina-.

ienal element at the. midpoint. of. the cycle portion which is a manifestation of the. non-agrea. inent oi theentered data. Itisseen, further, that. when, a denominational: control element, such as At (Fig. 6a.), is on eta mid-index. point time, it.

causes a related class, control element,. such as. A31, to shift also to on status, Finally. it. has. been shown that element, Ana. when. turned on, shifts.

elements Ba: and Or; to on state and element Bx, wh o r a l s oithe status. of element Ax, also shifts element Cm tolonstal e.

Reference to the tableoi illustrative data indi-.

cates that the remaining. classification data of frames 2 and 3 also disagree; i, e,, 8'76.5.l. and 56739.

Ct, and Cu are similarly operated upona condition of disagreement. in; the relatedorders. sufficient to state that such operation of each. element is a turning on and off of the element during acycle portionundier control of only. one of the non-matching designations of the related order... This. cycle portion. is the one which is indicated by the higher ofthe twn non-matching valueswhich may, be; either onthe-leading frame or the followin frame. At. the-midpoint of such- Ca to be turned. on. If the classelement has been turned on previously in the same cycle under control of onesuch tube related theretothe action of the other related tube-hasno; effect. theclass element simply. remaining, in on state: For in.-

stance, since element-.A.'tl isturned onunder cone. trolof tube SAL between. 9 and18?" of; cyclet In the manner, deseribedfor element. At, the other denomination elements Au, Bt, Be,

It .is.

causes the related classelement Ax, Bx, or-

22 the subsequent. action of tube sAu between the 8. and *7 times of the same cycle has no. effect on element Ase. However, if the element Arr had been off at the time the tube 3Au was rendered conductive, this tube would then have caused element A03. to turn on. Further, in the assumed example of data compared on frames 2. and 3', the element Ar wasthe first one turned on of all the class elements. In the manner explained before, as class element A3: is turned'on, it causes class elements "Br and Cr to be turned on, so that the control of the latter elements. by the related.

. denomination control tubes thereby is superseded;

When one or more of the class elements are turned on, they cause an interruption of record feeding operations in the following manner.

The point 58.5%?- of element Cr (Fig. 6b) is coupled by means of a condenser lfi'lbhb, resist.- ance ifit'ohb, and line ill (see also 6a) to point iiilio of element T. Element T has been turned on, in the manner described before in this; section so to energize the clutch magnets 6 .8. Condensers Hilhbb and resistance lfilllibb. correspond to. condenser liilb resistance its}; of Fig. 4. Hence, when the class element C50. is turned on, and its point I65?) rises in potential; a steep positive pulse is applied to point Nita of element T (Fig. 6a) shifting it to off status. is recalled that element Ca: (Fig. 679). became on at midway hetween.9 and 8 of cycle III. A.c.-

cordingly, T (Fig. 6a) is turned olf'at this time, as shown in Fig. 7., line (2 (111).

When element T is turned off, its pointv H35?) falls to; a low potential, thereby increasing: the grid bias of tube 546a, interruptingcurrent flow therethrough, and through the clutch magnets 53, causing their deenergization. Such deenergization of magnets lit. (see also Figs. 1 and 2) allows, the clutches. to latch up at D at the end of the cycle. Thus, the feed of film F is terminated with frames 3 and 4 about to pass the lower and upper sensing stations, respectively.

When a class control element, such as Ax (Fig. 6a), is on, its point this is at low potential. A

neon or like gaseous discharge lamp 2 l2, in serieswith a current limiting resistor 7H3 is connected between line 53d and point 555a. When this point. is at said low potential, the voltage differ- :encebetween the point and line is suflicient to ignite lamp 252, thus affording a visual manifestation that class element Ana is on; i. e., that there is a break in the major classifying data. The class elements Ba: (Fig. 6b) andCIr are also provided with neon lamps 2E2. which become ignited when said elements are on- In the exam: ple set forth in detailabove, all of the. class elements are on, and all thelamps 212: are ignited, visually manifesting breaks in the major, inltermediate and minor classifying data.

As long as element T. is oiT, clutch magnets 63 are deenergized and film feed suspended. At-the same time, one or more of elements Azr, Bar, and Ca." remains on, indicating the disagreement of control data. To restart film feed and turn off elements Ax, Br; and Cr, the operator reverses the position of switch 2% (Fig..6a.), causing condenser ['98 to discharge a steep positive pulse on wire 214. This pulse is applied via coupling condensers Ellen and related resistors to points [that of. elements A93, Brc, and Cr, turning them off.

When element Cr turns off, its point I851) falls in potential, so that a steep negative pulse is applied via condenser ltlbbb and resistor IBBbbband wire 2H. topoint itfia of element T. Element T, consequently, turns on and causes clutch,- mags

Images