US3774016A

US3774016A - Control of process according to registration indicia on material being processed

Info

Publication number: US3774016A
Application number: US00186234A
Authority: US
Inventors: R Sterns; R Mcguire
Original assignee: Sun Chemical Corp
Current assignee: Sequa Corp
Priority date: 1971-10-04
Filing date: 1971-10-04
Publication date: 1973-11-20
Anticipated expiration: 1990-11-20

Abstract

A web cut off device is controlled to produce cuts in accordance with registration marks on the web by measuring the deviation of each cut from its associated registration mark, inserting a first correction corresponding to this deviation and superimposing a second correction based upon the statistical processing of a number of preceeding deviation signals.

Description

United States Patent [1 1 Stems et al.

CONTROL OF PROCESS ACCORDING TO REGISTRATION INDICIA ON MATERIAL BEING PROCESSED Inventors: Robert B. Sterns, Great Neck;

Richard McGuire, Smithtown, both of N.Y.

Sun Chemical Corporation, New York, N.Y.

Filed: Oct. 4, 1971 Appl. No.: 186,234

Assignee:

[52] US. Cl 235/15l.1, 226/28, 340/259, 235/150.1

[51] Int. Cl. G061 15/46, B65n 23/18 [58] Field of Search 235/151.1, 151.13, 235/151.32, 92 DN, 92 EV; 250/219 LG; 340/259; 226/10, 28-30, 40; 101/226 References Cited UNITED STATES PATENTS l/1971 Crum 226/28 X 1 Nov. 20, 1973 3,585,372 6/1971 Bell et al. 235/92 EV 3,510,036 5/1970 Lewis, Jr. et al. 226/30 X 3,468,201 9/1969 Adamson et a1 226/29 X 3,525,858 8/1970 Thiede et al 226/28 X 3,601,587 8/1971 Thiede 226/28 X 3,397,634 8/ 1968 Betts et aL... 226/28 X 3,559,568 2/1971 Stanley 235/l51.1 UX

Primary Examiner-Eugene G. Botz Assistant Examiner-Jerry Smith Attorney-Cynthia Berlow [57] ABSTRACT A web cut off device is controlled to produce cuts in accordance with registration marks on the web by measuring the deviation of each cut from its asso ciated registration mark, inserting a first correction corresponding to this deviation and superimposing a second correction based upon the statistical processing of a number of preceeding deviation signals.

21 Claims, 7 Drawing Figures More. P0465 DEIAY 34 Car Lslvcm' AwusrME/vr CurDETEC 7-0,

PAIENTED NUY 20 1975 SHEET 1 EF 4 CONTROL OF PROCESS ACCORDING TO REGISTRATION INDICIA ON MATERIAL BEING PROCESSED This invention relates to process control and more particularly, it concerns novel methods and apparatus for maintaining proper registry between a machine which operates on a material in intervals and the material itself.

While different aspects of the present invention are applicable to many types of processing operations which occur at successive intervals, either in time or in space, on various media, the invention as a whole is particularly advantageous in web cut-off operations wherein a length of webbing, such as corrugated paperboard is cut into lengths corresponding to a preprinted pattern which repeates itself along the length of the web.

The present invention makes feasiable the preprinting of paperboard webbing prior to the use of that webbing as a facing for corrugated board which is thereafter creased and cut into predetermined lengths for later folding into containers and cartons of predetermined size. By printing on the webbing before it is glued to the corrugated board, a high quality image may be produced. However, once the board is printed, the locations where the board must be cut transversely into individual lengths become established; and unless the board is cut very close to those locations, the printed pattern will not be properly positioned on the panels of the finished carton or container which is later folded into shape from the cut lengths of board.

In general, a web of corrugated paperboard is severed into discrete lengths by moving it continuously between a pair of knife rolls. These rolls rotate in synchronism with each other; and knife blades on the rolls come together once during each revolution thereof to effect a transverse cut across the web. It will be appreciated that the location of a cut may be moved downstream, i.e., further along in the direction of board movement, either by speeding up the cutter roll rotation or by slowing down the speed of longitudinal web movement. Alternatively, the cut location may be moved upstream of web movement by either slowing down the cutter roll rotation or by speeding up the longitudinal web movement.

In order to maintain proper registry of the actual knife cut locations with the preprinted patterns on the web, the ratio of web speed to cutter roll speed must be controlled; and because of variations in the operating conditions, e.g., stretch in the web, slippage in web drive rolls, initial misadjustment, drift in the machinery and backlash or deadzone conditions in the machinery, adjustments must be made continuously to maintain proper registry.

In order to make these adjustments, it is necessary to ascertain the deviation of each actual knife cut location from a desired knife cut location. The present invention, in one aspect, provides novel techniques for obtaining representations of this deviation. According to the present invention, register marks spaced along the edge of the web in predetermined positional relationship with the printed pattern are detected by an optical scanner which produces a register mark signal whenever a web register mark passes the scanner. A knife cut detector is provided to produce knife cut signals whenever the cutter rolls produce a knife cut. ln addition, means such as a measuring wheel which rides on the web are provided to produce a series of closely spaced pulses as the web moves along so that each pulse corresponds to a given distance of web movement. The distance between the scanner and the location where a register mark should be when a proper knife cut is to be made is known; and this distance, as represented by a corresponding number of measuring wheel pulses, is preloaded into a counter. When a registration mark passes the scanner, the resulting register mark signal opens a gate which allows the measuring wheel pulses to be applied to a count-down terminal of the counter. The occurrence of a knife cut signal stops the count. If the knife cut signal occurs before the counter has counted down fully, the remaining count in the counter will correspond to the distance or deviation by which the actual cut was short of its proper location, so that the web element thus cutoff would be short. On the other hand, if before the knife cut signal occurs the counter counts down to zero, it will produce an output signal which will switch the application of subsequent measuring wheel pulses to a count-up input terminal of the counter so that these subsequent pulse will count 7 up from zero until the next knife cut signal occurs. The count remaining in the counter in this case corresponds to the distance or deviation by which the actual cut is long with respect to its proper location. In this case also, the output signal which switches the application of input pulses to the count-up terminal is also used to produce an indication that the deviation is long rather than short.

According to another feature of the present invention, error or deviation information relating to each operation of an intermittently operating process, for example cut location deviation in a web cut-off system, is processed in a novel manner to achieve an optimal correction situation. This feature of the invention involves first, applying the deviation information for one operation to the production of a corresponding adjustment for the next successive operation and second, applying the deviation information from the same operation to a statistical processing unit which processes this information with corresponding information from other operations to provide an average deviation correction signal. The average deviation correction signal is used to produce further correction which is applied to each of several successive operations. Thus, in the case of a web cut-off control, the measured deviation of each actual cut location from its corresponding cut locations is used to generate a corresponding correction to the system for changing cutter roll speed by a corresponding amount for the next subsequent cut. At the same time, this measured deviation is processed statistically with previously obtained deviation measurements and an average deviation correction signal is produced which is applied to change the cutter roll speed by a corresponding further amount for each of several succes sive cuts until further statistical processing produces a new or a changed average deviation correction signal.

The combining of individual deviation correction adjustments and average deviation correction adjustments serves to maintain close and accurate control, especially where the effect of each operation is dependent, to a certain extent, upon the deviation of a pre ceding operation. In the case of web cutting by knife rolls, for example, a cut made in the long direction of a registration mark will be followed by another cut tion of an average deviation for the system. In this manner, the effects of backlash or deadzone characteristics are cancelled by allowing them to distribute themselves substantially equally about a mean. In this way, the system may be accurately set according to any changes which may be present in the operating conditions.

The present invention provides improvements which, in certain circumstances effectively isolate the effects of backlash or deadzone characteristics from correctable deviations without requiring such a large number of successive measurements as would be necessary to have the effects of backlash or deadzone characteristics distribute themselves about a mean. These improvements involve combining the successive deviation indications algebraically and comparing each result of each combining operation with a different norm corre- I sponding to the number of operations undertaken since a previous average deviation correction was initiated. In the case of a web cutting system, for example, wherein it is known that the system is subject to backlash or deadzone effects in the amount of about onesixteenth inches, if the first deviation exceeds this amount, it can be safely assumed that more than backlash or deadzone effects are present in the deviation in dication. Accordingly, an average deviation correction can be generated immediately without need for several successive measurements. If two successive deviation signals, when combined algebraically, provide a total nearly twice the amount of any deadzone or backlash effects, then again an average deviation correction signal may be generated. Similarly, as successive deviations are combined, the resulting combination is compared with a different norm, and if such combination exceeds the norm, a correction signal will be generated.

There has thus been outlined rather broadly the more important features of the invention in order that the detailed description thereof that follows may be better understood, and in order that the present contribution to the art may be better appreciated. There are, of course, additional features of the invention that will be described hereinafter and which will form the subject of the claims appended hereto. Those skilled in the art will appreciate that the conception upon which this disclosure is based may readily be utilized asa basis for the designing of other structures for carrying out the several purposes of the invention. It is important, therefore, that the claims be regarded as including such equivalent construction as do not depart from the spirit and scope of the invention.

A specific embodiment of the invention has been chosen for purposes of illustration and description and is shown in the accompanying drawings, forming a part of the specification, wherein:

FIG. l is a schematic diagram of a web cut-off system in which the present invention is embodied; 65

FIG. 2 is a plan view showing a length of preprinted web which may be cut into lengths with the system of FIG. 1;

FIG. 3 is an enlarged view taken along line 3-3 of of FIG. ll showing cutter rolls with knife blades in their relative positions just prior to making a cut;

FIG. 4 is a view similar to FIG. 3 showing the knife blades in their relative positions at the execution of a cut;

FIG. 5 is a graph illustrating relative rotational velocities and positions of the cutter rolls of FIGS. 3 and 4; and

FIGS. 6A and 68 together form a block diagram of a correction unit which is used in the system of FIG. 1.

The registration controlled web cutting system of FIG. I operates to cut a preprinted paperboard web 10 into elements 12. The web 10 is printed with a recurring pattern; and it is important that each element 12 contain the pattern properly centered or registered thereon so that the element can later be folded into a carton or other packaging arrangement with the printed pattern properly displayed thereon. The present invention serves to ensure that the cutting of the web 10 is such that the length of each element 12 is properly related to the printed pattern so that each element will contain the pattern accurately centered thereon.

As shown in FIG. 1, the web 10 passes through a pair of drive rolls 14 and is driven thereby in the direction of an arrow A. The web 10 then passes a set of slitterscorer rolls 16 which form longitudinal crease lines and slits for later folding of the web into a carton configuration. After passing through the slitter-scorer rolls 16, the web 10 passes between a set of cutter rolls 18 which sever the web transversely into discrete elements 12.

As can be seen in FIG. 2, the web 10 is printed with a recurring pattern P. Each pattern is accompanied by a registration mark 20 located along one edge of the web. These registration marks have a preselected positional relationship with the pattern and with a corresponding line of cut 22 (shown in dashed line) where ces sive pattern repeats and successive lines of cut 22),

is indicated as]. in FIG. 2.

Reverting now to FIG. I, it will be seen that there is provided a common drive motor 24 which is connected to drive the drive rolls 14, the slitter-scorer rolls l6 and the cutter rolls I8 in synchronism. The drive train be tween the motor 24 and the cutter rolls 18 includes a differential transmission 26, a Reeves mechanism 28 and a cycle cut mechanism 30. The differential transmission 26 and the Reeves mechanism 28 both function to control the ratio of cutter roll to drive motor rotation. The cutter rolls each contain a knife blade 32 term adjustments since it requires a certain amount of time following each adjustment to reach its normal operating condition for the adjustment. Also, the Reeves mechanism is generally unsuitable for fine adjustments.

Because of these limitations, the Reeves mechanism is used primarily for presetting the system prior to operation thereof, while the differential transmission 26 is used to make ratio changes i.e., length of cut corrections, during operation of the system.

The manner in which the differential transmission is controlled will be described more fully hereinafter.

The cycle cut mechanism 30 serves to modify the rotation of the cutter rolls 18 so that their velocity in the vicinity of their coming together for cutting is slightly greater than that of the web. This serves to improve the sharpness of the cut and helps to prevent torn or ragged edges on the various cut elements 12. This is exemplitied in FIGS. 3 and 4. In the example shown in FIG. 3, the web 10 moves in the direction of the arrow A, while the knife blades 32 are carried around through the are 6, by their respective rolls 18. Thereafter, as shown in FIG. 4, the knife blades 32 traverse the are during which they sever the web 10. During the first half of the arc 6 the knife blades undergo an increase in velocity up to the point of actual cut; and thereafter the knife blades undergo a decrease in velocity to the end of the are 0 whereupon they continue to move at a lower velocity throughout the arc 0 This movement of the knife blades is illustrated diagramatically in FIG. 5 wherein cutter roll rotational velocity is plotted against cutter roll rotation. As can be seen, the cutter rolls 18, and their associated knife blades 32, move at lower velocity throughout the arc 6 and they undergo an increase followed by a decrease in velocity while traversing the are 0 This occurs once during each 360 degrees of rotation of the cutter rolls.

The length of cut i.e., the amount'of web that moves between the cutter rolls 18 between successive knife cuts is determined by the average linear speed of the web and the average rotational speed of the cutter rolls 18. By decreasing the average cutter roll speed for a given web speed, the length of cut can be increased, and by increasing the average cutter roll speed, the length of cut can be decreased. It will be appreciated from FIG. 5 that the average cutter roll velocity for each complete rotation thereof will be slightly greater than the constant velocity throughout the arc 0,. As in dicated previously, this average velocity may be preset with the Reeves mechanism 28 and it may be adjusted during operation of the system by means of the differential transmission 26.

Adjustment of the differential transmission 26 is made, as shown in FIG. 1, by the rotational movement of a planetary carrier 34 between input and output gears 36 and 38. A plurality of planet gears 40 are carried by the planetary carrier 34 and are meshed with each of the input and output gears 36 and 38. For a given rotational speed of the drive motor 24 and the input gear 36, the speed of the output gear can be increased or decreased by an amount corresponding to the direction and speed of rotation of the planetary carrier 34. This, in turn, is controlled from a correction unit 42.

The correction unit 42 receives registration pulses from a registration scanner 44 located near the cutter rolls 18. A cut detector 46 is also positioned adjacent the cutter rolls and is connected to produce a knife pulse upon each severing operation of the knife blades 32. These pulses also are applied to the correction unit 42. Finally, a measuring wheel 48 is mounted to ride on the web 10 and to turn as the web advances. The measuring wheel is coupled to a pulse generator 50 which operates to produce a web movement pulse for wheel rotation corresponding with each hundredth of an inch of web movement in the direction of the arrow A. These web movement pulses also are applied to the correction unit 42.

OPERATION OF OVERALL SYSTEM During operation of the system of FIG. 1, the drive rolls l4 advance the web 10 toward and between the cutter rolls 18. At the same time, the cutter rolls are rotated and their knife blades 32 undergo a cyclic pattern of movement described above in connection with FIGS. 3, 4 and S, with the average rotational speed of the cutter rolls l8 governing the length of each cut made.

As each cut is made, a measurement is made to ascertain the deviation of the actual cut length from the desired cut length. This measurement is used in the correction unit 42 to produce adjustments of the differential transmission 26 to obtain control of succeeding cuts.

Essentially, the measurement of deviation of each actual cut from each desired cut are made by obtaining an indication of the amount of web movement which takes place between the time that a registration mark 20 is detected by the registration scanner 44 and the time the next subsequent knife pulse is produced by the cut detector 46. In the case illustrated, the registration mark 20 is shown to be positioned along the lines of cut 22. Thus, for the actual knife cuts to take place on the lines of cut 22, the web 10 should move a distance equal to the distance between the registration scanner 44 and the point of cut of the knife blades 32 during the time between the occurrence of a pulse from the registration scanner 44 and the occurrence of a knife pulse from the cut detector 46. This movement, as pointed out above, is indicated by the action of the measuring wheel 48 and the pulse generator 50.

It is not necessary that the registration marks be in exactly the same position as the desired lines of cut 22. In the event that they are offset, the amount of this offset can be added to or subtracted from the distance the web is to move between the detection of a registration 42 which processes this information and makes appropriate adjustments to the differential transmission 26 which increases or decreases the average speed of the cutter roll rotation so that subsequent cuts by the knife blades 32 will be brought into closer registration with corresponding desired lines of cut 22.

THE CORRECTION PROBLEM Because of certain characteristics of the abovedescribed processing, i.e., web cut-off system, a simple feedback control arrangement is not feasible for obtaining close and accurate control of output.

The first characteristic lies in the cut-off machinery.

Because of several factors, including the peculiarities of the Reeves mechanism 28 and the fact that the cycle cut mechanism 30 causes the knife blades to undergo an acceleration and a deceleration during each cutting operation, a certain region of randomness exists due to backlash in gear systems and various dead zones in the different mechanisms. It is, of course, impractical to attempt to control any machine with any greater degree of precision than the dead zones present in the machine will allow. For example, if the backlash in the system herein will allow the knife blades 32 to be moved by hand by a certain amount, e.g., i l/ 16 inch, without any turning of the shaft of the drive motor 24, then it is impractical, with any kind of control, to obtain a degree of precision which will ensure that all pieces cut will deviate from a desired length by less than onesixteenth inch.

The second characteristic lies in the nature of the cut-off process wherein the length of a subsequent cut is governed not merely by the setting or operation of the cut-off machinery, but also by the location of a previously made cut. Thus, if the distance between successive registration marks 20 along the web is, say, 3 feet, and if the machinery deviates to a cut length of 2 feet, eleven inches so that a preceding cut is made off register by one inch, then, even if the machinery is corrected by one inch to make a cut length of 3 feet, the next subsequent cut will also be 1 inch off from its associated register mark, since the 3 feet is measured from the previous cut which, itself, was 1 inch off register. Now if the system is corrected to make a cut length of 3 feet 1 inch, then the next subsequent cut will be on register; however, following cuts will be off register in the opposite direction.

It will be appreciated that an over correction is necessary to bring the system back to registry. However, as pointed out, an overcorrection causes misregistration in the opposite direction of following cuts. Such an arrangement has a tendency toward instability since the corrections, in order to be effective, must be such as to produce comparable misregistration in the opposite direction.

tinued correction or system adjustments are made based upon the results of the statistical processing. The application of a phase correction eradicates the effects of a preceeding misregistered cut while the statistical processing of the error or misregistration measurement permits a proper adjustment of the system which is sufficient to correct any errors without however overcorrecting to the point of instability.

The statistical processing of detected misregistrations to eliminate their deadzone or backlash components is based upon the realization that these components are purely randomly distributed and therefore over a large number of successive measurements they will cancel themselves out. In the present case, the misregistration errors for at least five successive cuts are combined algebraically. Should the cumulative algebraic sum of five or more successive misregistration errors exceed a given minimum number then that sum is taken as the adjustment error of the system and a corresponding correction, amounting to a system adjustment, is made. In the present case, this system adjustment is made by superimposing on the phase correction for each cut, a system or average deviation correction. While the phase correction is made only once however, the system or average deviation correction is repeated for each cut until further statistical processing causes it to change.

The present invention additional provides refinements to the statistical processing whereby, under certain circumstances, it serves to separate the deadzone or backlash components from system adjustment errors in less than five measurements. This is achieved by comparing the magnitude of these initial misregistration errors with certain criteria setup in accordance with the known system dead zone characteristics. Thus, if the first measurement made indicates a misregistration of say 3/32 inch while the system dead zone is l/ 16 inch, then it may be assumed from this one measurement that a system error exists which is at least The instability problem is further complicated by the precision limitations of the system described previously. This is because it is not possible to know, on an individual measurement which shows a cut misregistration of say plus three thirty-seconds inch, how must of this misrepresentation is attributable to dead zone or backlash variation in the machine and how much is due to improper adjustment of the system. If the 3/32 inch error is made up of 1/16 inch backlach or dead zone error and only l/32 inch system misadjustment, then any correction which is made with the assumption that the entire error is system misadjustment will upon correction, introduce an oppositely directed adjustment error of more than 1/16 inch, thereby causing instability. On the other hand, if any correction is made with the assumption that the entire error is dead zone error then the system will never be brought back to full registration.

The present invention handles the above problem by using each measurement made in two ways: first, the

measurement of misregistration is used to make a phase correction, that is a correction in the opposite direction from the misregistration, for the next subsequent cut only, and second, the measurement of thisregistration is statistically processed to eliminate its deadzone or backlash components and thereafter con- 1/32 inch. Further, if upon two successive measurements, the cumulative algebraically combined misregistration error amounts to 1% inch, it may be assumed that a system error exists.

The present invention thus statistically processes successive misregistration measurements and separates their system misadjustment components from the dead zone components by combining several successive measurements algebraically.

THE CORRECTION UNIT As shown in FIG. 1, the correction unit 42 includes a measurement section 52, a program control section 54, an average length correction section 56, an average length accumulation section 58 and a correction section 60.

Signals from the registration scanner 44, the measuring wheel pulse generator 50, and the cut detector 46 are supplied via a registration pulse line 62, a digital wheel pulse line 64 and a knife pulse line 66 to the measurement section 52. Knife pulse signals are also supplied to a delay circuit 68 and delayed knife pulses are applied to the program control section 54.

The various signals supplied to the correction unit 42 are processed therein to provide measurements of cut misregistration. These measurements are processed, as described above, individually to provide phase corrections, and statistically to provide system or average length corrections. These two types of correction are combined in the correction section 60 and are converted therein to a corresponding mechanical shaft rotation. This rotation is transmitted (as indicated by a heavy dashed line 70) to the planetary carrier 34 of the differential transmission 26 for adjusting the average speed of rotation of the cutter rolls 18 which, in turn, controls the length of web cut between successive operations of the knife blades 32.

The various sections of the correction unit 42 will now be described in detail.

THE MEASUREMENT SECTION The purpose of the measurement section is to obtain a count indicative of the distance along the web by which each actual knife cut deviates from its desired location. In general, this is achieved by first storing in an up-down counter, a count corresponding to the distance a registration mark should have travelled from the registration scanner 44 at the time a knife cut is made. When a registration mark is detected, the count is loaded into the up-down counter 72, and then as the web 10 moves along the pulses from the measuring wheel pulse generator 50 (which correspond each to 0.01 inches) are applied to reduce the stored count. Ideally, the count should be zero at the time a knife pulse, which stops the count, occurs. Should the knife pulse occur before down is complete, the remaining count will indicate the amount by which the cut length is too short; and should the knife pulse not occur until after count down is complete, the counter will count up again and produce a long indication of the amount by which the actual cut length is too long.

.The measurement section transfers these long and SHORT INDICATIONS DIRECTLY TO THE COR- RECTION SECTION FOR PHASE (i.e., cut to cut) correction, and to the average length correction section for statistical processing.

Referring now to FIG. 6A, the measurement section 52 of the correction unit 42 is shown to include a series of decade counters 72. These decade counters are constructed to count applied pulses according to a binary decimal code. The counters 72 are also adapted to be preset manually to any desired count, as indicated by double arrows 74 extending up to each counter. The counters represent, respectively, hundredths of an inch, tenths of an inch, inches and tens of inches. Each applied pulse represents one one hundredth of an inch. The counters will count up or down and thereby increase or decrease their count in response to pulses applied to count up and count down

terminals

76 and 78 respectively.

A count storage register 80 is arranged in association with the decade counters 72 to receive and retain the information present at certain times in the decade counters 72. A digital display arrangement 82 is also provided and, as shown by the double arrows leading thereto from the count storage register 80, is adapted to receive, for display purposes, the count present in' the storage register 80. There are also provided preset AND gates 84 which receive information from the storage register 80 and compare it to preselected criteria The count-up terminal 76 of the decade counters 72 is connected to receive signals from an AND-NOT gate circuit 92. This circuit has an inhibit terminal 94 connected to receive input signals from a short output terminal 96 of a direction flip-flop circuit 98. The remaining input terminal of the AND gate circuit 92 is connected to the output terminal of a digital wheel pulse AND gate circuit 100. One input of the digial wheel pulse AND gate circuit 100 is connected to the digital wheel pulse line 64 for receiving input pulses representative of web movement. The other input terminal of the digital wheel pulse AND gate circuit 100 is connected to a start count output terminal 102 of a start measurement flip-flop circuit 104. The corresponding input terminal of the start measurement flipflop circuit 104 is connected to the registration pulse line 62 to receive registration pulses from the registration scanner 44. The other input terminal of the start measurement flip-flop circuit 104 is connected to the knife pulse line 66 to receive knife pulses from the cut detector 46. short indications directly to the correction section 60 for phase (i.e., cut to cut) correction, and to the average length correction section for statistical processing.

One of the inputs to the short measure AND gate circuit 108 is applied from the digital wheel pulse AND gate circuit 100, while the other input terminal of the short measure AND gate circuit is connected to receive inputs from the short output terminal 96 of the direction flip-flop circuit 98. The direction flip-flop circuit 98 is also provided with a long output terminal 110 which is energized upon the reception of signals at a long input terminal 112. The long input terminal v 112 is connected to receive signals from a borrow outfor producing signals on an accept cut line 86, and I unaccept cut line 88 and over four inch line 90. These lines are connected to various display and control circuits (not shown) for providing a continuous indication of the operating conditions of the system.

put terminal 1 14 of the decade counters 72. Signals appear at this borrow output terminal 114 whenever the decade counter passes through a zero count in the down direction.

The short" output terminal 96 of the direction flipflop circuit 98 is connected to the count storage register to indicate the direction, i.e., short or long or error represented by the numbers which are transferred to the storage counter. These signals are supplied from the storage counter via a plus-minus line 1 16 to the digital display 82. The count storage register 80 has a storage input terminal 1 18 connected to the knife-pulse line 66 so that it will accept and store information present in the decade counters 72 upon the reception at the terminal 118 of a knife pulse signal.

The presetting of the decade counters 72, as indicated previously, is undertaken manually prior to operation of the system. The actual preset count, however, is not transferred to the decade counters until the reception of a load signal via a decade counter load line 120. This line, as shown in FIG. 6A, is connected to the registration pulse line 62.

THE PROGRAM CONTROL SECTION The program control section 54 serves to control the sequence of operation of the various sections of the correction unit 42. As shown in FIG. 6A, the program and are applied to the multiplex unit 124. The multiplex unit 124 produces output signal commands on various circled command lines indicated by the numbers 1 through 7. Whenever the demultiplex unit receives a signal on one of its arrowed boxes of a given number, it causes a signal to be supplied via the switch line 126 so as to produce output commands on the circled command line of the next higher number. The switching sequence is initiated by a program counter 128 which maintains sequence of switching and which supplies clock pulses at the various command output terminals. The program counter 128 receives input signals from a command AND gate 130 and this, in turn, receives input signals from a source of clock pulses (not shown) via a main clock pulse line 132, demultiplex signals from the switch line 126 and program start signals via a program start line 134 from a program start flip-flop 136. The program start line 134 is energized by application of a delayed knife-pulse signal to a start terminal 138 of the program start flip-flop 136. The other input terminal of the program start flip-flop circuit 136 is connected to a program stop OR gate 140 which receives program stop input signals via a register mark line 142, a preprint automatic line 144 and the over four inch line 90 from the measurement section 52. It will be noted that in the various sections of the correction unit 42, there are shown arrowed boxes numbered 1, 3 and 7. Whenever signals appear on these boxes, they are applied simultaneously to correspondingly numbered boxes indicating input to the demultiplex unit 122. Similarly, signals which are generated at the command lines of the multiplex unit 124 are applied to corresponding circled linessimilarly numbered in the various sections of the correction unit. ln addition, the demultiplex unit 122, the multiplex unit 124 and the program counter 128 are set to advance automatically, without inputs to the arrowed boxes numbered 2, 4 and 6, respectively, to advance the outputs of the multiplex unit 124 so that command appear on the circled

command lines

3, 5 and 7.

In operation of the program control section 54, the occurrence of a delayed knife pulse signal from the knife pulse line 66 is applied to the program start flipflop circuit 136 causing it to produce a signal on the program start input line 134. This signal opens the command AND gate 130 causing it to apply clock signals to the program counter 128. The program counter 128 first causes clock pulse signals to occur at the No. 1 circled command line of the multiplex unit 124. Corresponding circled command lines No. l are found in the measurement section 52 and in the correction section 60, and the clock pulses are applied to both these lines simultaneously. These clock pulses continue until a signal appears atthe borrow output terminal 114 of the decade counters 72 in the measurement section, indicating that these counters have been reduced to a zero count. Since the terminal 114 is shown connected to an arrowed box No. l a call-back signal is applied at this time to the correspondingly numbered arrowed box in the demultiplex unit 122. This signal is processed in the demultiplex and multiplex units so that the pulses on the circled command line No. 1 stop and a signal instead appears at the circled command line No. 2. As can be seen in the drawing, and as'will be explained more fully hereinafter, this signal is applied to the average length accumulator section. The signal on the No. 2 command line is automatically terminated by the program counter 128 and the multiplex unit 124 after a predetermined duration, and the multiplex unit is advanced to produce a series of clock pulse signals on its circled command line No. 3. As shown in the drawings and as will be described more fully hereinafter, these clock pulse signals are applied simultaneously to correspondingly numbered circled command lines in the average length correction section 56 and in the average length accumulator section 58. The clock pulses continue on the No. 3 circled command lines until a call-back signal appears at a call-back line indicated by the arrowed box No. 3 in the average length correction section 56. This call-back signal also appears at the call-back line indicated by the arrowed box No. 3 at the demultiplex unit 122 in the program control section 54. This signal is processed in the control section and causes the multiplex unit 124 to terminate the clock pulses on the circled command line No. 3 and instead to produce a command signal on the circled command line No. 4. This command signal is applied to a corresponding circled command line No. 4 in the average length accumulator section 58.

The program counter 128 automatically terminates the signal on the No. 4 circled command line after a preselected time and then causes the multiplex unit 124 to produce a series of clock pulses on a circled command line No. 5. As shown in the drawings, these clock pulses are applied simultaneously to corresponding circled command lines No. 5 in the average length accumulator section 58 and in the correction section 60.

The clock pulses on the No. 5 circled command lines continue until a call-back signal appears at the callback line indicated by the arrowed box No. 5 in the average length accumulator section 58. This signal is applied to the demultiplex unit 122 in the program control section 54 causing it to advance the multiplex unit 124 so that the clock pulse signals are terminated and a signal appears for a preselected duration on the No.

nal is applied to a corresponding line in the average length accumulator section 58.

After the preselected duration, the signal on the No. 6 circled command line is terminated and the program control section automatically advances to cause a series of pulse signals to occur at the N0. 7 circled command line. These signals are applied to the correction section 60 until a call-back signal appears at a call-back line indicated by the arrowed box No. 7 in the correction section. The call-back signal is applied to the demultiplex unit 122 and terminates its operation and resets it to undergo a similar sequence of events following a subsequent knife cut.

THE AVERAGE LENGTH CORRECTION SECTION tion differs from the phase correction in that it is repeated for all subsequent cuts and is changed only by the amount of a subsequently generated length setting correction. The phase correction, on the other hand occurs only for the knife cut which immediately follows the generation of each phase correction signal.

The statistical processing of the deviations of several successive cuts consists essentially in algebraically combining the successive deviations and then producing a correction signal when the mean of a large number of thsee individual deviations exceeds a predetermined limit. This serves to nullify the random effects of backlash and dead-zone conditions in the system since the average or mean of a large number of successive random deviations within the dead-zone should occur in the center of the zone.

The large number of these successive measurements has, in the illustrated case, been chosen empirically as five; and the mean deviation beyond which an average length setting correction is made is also empirically taken at plus or minus 0.32 inches for a system wherein a total deadzone of 0.12 inches exists.

The statistical processing of several successive deviation indications is further modified, according to the present invention, by monitoring the extent of the algebraically combined deviations for less than five and by producing correction signals whenever the combined deviations exceeds predetermined limits corresponding to the number of cuts which have been combined since a previous correction. This provides rapid length correction for situations where substantial adjustments must be made to the system. For example, where it is known that the deadzone region within which the system may vary its cut deviation for a given setting is a total of 0.12 inches, and the initial cut has deviated off register by 0.13 inches then it can be said that there is a substantial error in the system setting and an adjustment should be made. In such case to wait until five successive cuts have been made is not necessary.

In the illustrated embodiment the following criteria for system correction are established in the average length correction section:

Accumulated Algebraic Deviation Which, if

The average length correction section 56 includes a two decade counter 150 having an up-count input terminal 152 and a down-count input terminal 154. The two decade counter 150 also includes a borrow terminal 156 which produces a signal each time the counter reaches a zero count. A decoder circuit 158 is provided to receive count information from the two decade counter 150 and to transfer this information to various AND gate circuits 159. The decoder circuit 158 presents signals upon various output terminals 160 corresponding to the count in the counter circuit 156. As shown in the drawing, the output terminals used in the present case are those which represent counts corresponding to 13, 24, and 32 one hundreths of an inch, respectively. It is these terminals which are connected to corresponding ones of the AND gate circuits 159. Each of the AND gate circuits 159 is also connected to receive signals from associated ones of output terminals 162 of a knife pulse counter 164. The knife pulse counter 164, as indicated, has a five count capacity. Knife cut signals from the knife pulse line 66 are applied to the knife pulse counter 164, and upon the reception of successive knife pulse signals, successive ones of the output terminals 162 become energized. The last of the terminals, i.e., that indicated by the numeral 5, remains energized for every pulse following the fifth pulse unit] a reset signal is applied to a reset terminal 166 thereon. The outputs of the various AND gate circuits 159 are connected to corresponding loading input terminals 168 of a correction counter 170. As

indicated, the output of the AND gate circuit 159, which is connected to the first output terminal 162 of the knife pulse counter, is connected to the loading input terminal 168 of the correction counter corresponding to a count of five and thereby a signal from this AND gate circuit loads a five count into the correction counter 170.

The next two AND gate circuits 159, which are connected, respectively, to receive input signals from a second and third output terminals 162 of the knife pulse counter 164 and from the 24 count output terminal 160 of the decoder circuit 158, are applied via a first OR gate circuit 171 to the correction counter input terminal 168 corresponding to a four count to be loaded into the correction counter. The remaining AND gate circuits 159 are connected to the correction counter input terminals 168 corresponding to a three count and a one count respectively. The correction counter 170 additionally is provided with a load terminal 172, which, upon the reception of a proper command signal, causes the correction counter to be loaded with a count corresponding to the particular one of its load input terminals 168, which happens to be energized. Thereafter, countdown signals are applied via one of the command lines to an input terminal 174 which cause the count which has been loaded into the correction counter 170 to diminish. Upon complete removal of the count from the correction counter 170 in this manner, the correction counter will produce a signal on a borrow output terminal 176 and, at the same time, will send a callback signal back to the demultiplex unit 122 of the program control section 54.

The up-count input terminal 152 of the two-decade counter 150 is connected to receive input pulses from an AND circuit 178, while the down-count input terminal 154 is connected to receive input signals from a AND-NOT gate circuit 180. One input terminal of each of the AND and AND-

NOT gate circuits

178 and 180 is connected to the No. 1 circled command signal line from the multiplex unit 124 in the program control section 54. The inhibit terminal 182 of the AND-NOT gate circuit and the remaining input terminal of the AND gate circuit 178 are both connected to receive input signals from an EXCLUSIVE OR gate circuit 184. This EXCLUSIVE OR gate circuit 184 receives long signals from the long output terminal of the direction flip-flop circuit 98 in the measurement section 52. The remaining input terminal of the EXCLUSIVE OR gate circuit 184 is connected to receive outputs from one side of a second direction flip-flop circuit 186. The direction flip-flop circuit 186 has a common input terminal 188 which is connected to receive borrow signals from the borrow input terminal 156 of the two-decade counter so that each time the two-decade counter 150 has its count reduced to zero, a signal is applied to the common input terminal 188 of the direction flipflop circuit 186, causing the output terminal thereof connected to the EXCLUSIVE OR gate circuit 184 to change state.

tions for a plurality of knife cuts along the web 30 and to provide a length setting correction to the system corresponding thereto. Thus, for each deviation measurement which appears in the decade counters 72 of the measurement section 52, this deviation amount is transferred to the two-decade counter 150 and is algebraically combined with counts applied to the twodecade counter 150 from previous measurements. The algebraic combining of successive deviation indications from successive measurements is made by application of clock pulses via the No. 1 circled command line to the AND and AND-

NOT gate circuits

178 and 170. If no input signal is being applied to the inhibit terminal 182 of the AND-NOT gate circuit 180, the clock pulses from the No. l circled command line are applied through the AND-NOT gate circuit 180 to the downcount input terminal 154, thereby counting down toward zero in the counter. If, during this countdown, the number in the two-decade counter 150 becomes reduced to zero, a signal will appear on its borrow terminal 156 and this signal will be applied to the input terminal 188 of the direction flip-flop circuit 186. As a result, a signal is produced at the output of the EX- CLUSIVE OR gate circuit 184 and is applied to the inhibit terminal 182 of the AND-NOT gate circuit 180. This signal is also applied to the second input terminal of the AND gate circuit 178. As a result, the following clock pulse signals from the No. l circled command line are applied through the AND gate circuit 178 to the up-count input terminal 152 of the two-decade counter 150, thereby causing its count to increase. Thus, for example, if the two-decade counter 150 has a count in it of five, and then ten successive clock pulses are to be transferred from the decade counters 72 of the measurement section 52, these clock pulses will first be applied via the AND-NOT gate circuit 180 so that the count in the two-decade counter 150 becomes reduced to zero after the occurrence of five of these clock pulses. The remaining five clock pulses, however, will be applied to the up-count input terminal 152 for a net count of five, which is equal in magnitude to the algebraic sum of the newly added count and the previous count in the counter; It will be appreciated that the two-decade counter 150 in the described system always counts in an upward direction from zero and, at the same time, serves to provide a count whose magnitude corresponds to the algebraic sum of the previously supplied signals.

As each measurement is made and transferred to the two-decade counter 150 as above-described, another knife pulse signal is applied via the knife pulse line 66 to the knife pulse counter 164. If, after a first knife pulse and the measurement associated therewith, the decoder 158 produces a signal at its output terminal 160 corresponding with a deviation of 0.13 inches, then both inputs of the AND gate circuit 159 associated with the five count input terminal 168 of the correction counter 170, will be energized and a five count will be loaded into the correction counter. However, if the first measurement shows a deviation of less than 0.13 inches no count will be loaded into the correction counter. If,

however, on the second measurement the magnitude of the algebraically combined first and second measurements should produce an indication corresponding to count output terminal 162 of the knife pulse counter 164, will be energized and will produce an output signal which will be applied via the OR gate circuit 171 to the four count input terminal 168 of the correction counter 170, thereby loading a count of four into the correction counter 170. In a similar manner, other counts may be loaded into the correction counter 170, corresponding to measured deviations,

After each measurement is made, a signal is applied via the No. 2 circled command input line to the load terminal 172 of the correction counter 170, allowing the correction counter to become loaded to the count corresponding to the particular one of its input terminals 168 which is energized. Following this loading, a

series of clock pulses appears upon the No. 3 circled command line and these are applied to a count-down input terminal 174 on the correction counter. If no count has been placed into the correction counter 170,

then immediately upon the application of the first countdown pulse, a signal will appear on the borrow line and will produce a callback signal on the No. 3

call-back signal line of the demultiplex unit 122 to advance the program. On the other hand, if a finite count is present in the correction counter 170, the application of clock pulses to the correction counter 170 will continue until the correction counter reaches zero, whereupon a signal will appear on its borrow line, and on the No. 3 callback signal line.

It will be appreciated from the foregoing that as each measurement is made, the measurement count is transferred from the decade counters 72 in the measurement section 52 to the two-decade counter 150 in the average length correction section 56 and is there combined algebraically with those counts which have previously been entered and previously been combined algebraically with each other in the two-decade counter. Whenever this algebraic total exceeds an amount corresponding to the count in the knife pulse counter that total is loaded into the correction counter. If no such loading takes place prior to the knife pulse counter reaching a count of five, then the knife pulse counter will remain at a five count for subsequent measurements and these measurements will continue to be rection signals. The resulting average length correction is thereafter transferred to the correction section where it is combined with a phase correction count and converted to a shaft rotation for adjustment of the differential transmission 26.

Whenever a count is loaded into and transferred from the correction counter 170 the knife pulse counter 164 and the two-decade counter are both reset to zero so that deviation measurements for knife .cuts subsequent to resulting average length correction may be statistically processed and, if necessary a subsequent average length correction may be similarly derived.

THE AVERAGE LENGTH ACCUMULATOR SECTION The purpose of the average length accumulator section is to receive and store the count which was generated in the last previous operation of the correction counter 170 in the average length correction section 56. The signal from the correction counter 170 is removed from the correction counter following operation of the average length correction section 56 so that it can begin a new statistical processing of subsequent measurements. The average length accumulator section 58 serves to combine algebraically the average length correction signal from the correction counter 170 with an average-length correction signal previously generated and presently being applied to each knife cut operation, thereby updating the average length correctionJThe average length correction which is thus derived is transferred, for knife cut made, to the correction section for combination with a phase correction and from there to the differential transmission 28 for adjustment of the average rotational speed of the cutter rolls 18. I

As shown in FIG. 6B, the average length accumulator section 58 includes a two-decade counter 200 and a storage register 202. As indicated by a first set of double arrows 203, signals from the two-decade counter may be transferred and stored in the storage register 202. This will occur upon the application of an input signal to a load input terminal 204 whenever the No. 4 circled input command line is energized. Similarly, the count in the storage register 202 may be transferred back, as indicated by a, second set of double arrows 205, upon the application of a reload signal to a reload input terminal 206 upon energization of the No. 6 circled input command line.

The two-decade counter 200 is provided with an upcount input terminal 208 and a down-count input terminal 210. The up-count input terminal 208 is connected to receive input signals from an AND gate circuit 212. One terminal of the AND gate input circuit 212 is connected to receive clock pulses from the No. 3 circled input command line and, at the same time, to receive input signals from an EXCLUSIVE OR gate circuit 214. The down-count input terminal 210 is connected to receive signals via an OR gate circuit 216 from either and AND-NOT gate circuit 218 or from the No. circled command line. The AND-NOT gate circuit 218 has an inhibit terminal 220 connected to receive outputs from the EXCLUSIVE OR gate circuit 214. The remaining input terminal of the AND-NOT gate circuit 218 is connected to receive clock pulses from the No. 3 circled command line. One input terminal of the EXCLUSIVE OR gate circuit 214 is connected to receive long input signals from the long output terminal 110 of the direction flip-flop circuit 98 in the measurement section. The other input terminal of the EXCLUSIVE OR gate circuit 214 is connected to receive input signals from one side of a direction flipflop circuit 222. The direction flip-flop circuit 222 contains a common input terminal 224 which is connected to receive signals from a borrow output terminal 226 of the two-decade counter 200. The borrow input terminal 226 is also connected to supply signals via the No. 5 callback line to the No. 5 input terminal of the demultiplex unit 122 in the program control section 54.

A reset terminal 228 is provided in the two-decade counter 200 for resetting the counter to zero at the start of operation. This signal may be applied manually to clear the two-decade counter 200.

In operation of the average length accumulator section 58, signals are applied from the No. 3 circled command line to one of the

input terminals

208 or 210 of the two-decade counter 200. This occurs simultaneously with the counting down of the count in the correction counter 170 so that the count in that counter is effectively transferred to the tWOdCCadC counter 200 in the average length accumulator section 58. When this transfer is completed and a signal appears at the borrow terminal 176 of the correction counter 170, that signal is also applied to the No. 3 callback line. This causes the program control section 54 to advance its operation and terminate the pulses on the N0. 3 circled command line and atthe same time, to produce a signal for a predetermined duration on the No. 4 circled command line. This last-mentioned signal, as indicated in the drawing, is applied to the load input terminal 204 of the storage register 202, thereby causing the count in the two-decade counter 200 to produce an identical count in the storage register 204. When this transfer has been completed, the program control section 54 automatically terrninates the load signal on the No. 4 circled command line and causes a series of clock pulses toappear on the No. 5 circled command line. These clock pulses are applied through the OR gate circuit 216 to the down-count terminal 210 of the twodecade counter 200. This down-count continues until the two-decade counter 200 reaches a zero count at which time a signal appears on its borrow output terminal 226 and on the No. 5 callback line. The signal on the No. 5 callback line is applied to the program control section 54 and causesit to advance its program to terminate clock pulses on the No. 5 circled command line and instead, to produce a reload pulse for a short duration on the No. 6 circled command line. The

counting down of the two-decade counter occurs simultaneously with the application of clock pulses to the correction section 60. This serves to transfer the count present in the two-decade counter 200 to a counter in the correction section 60. When this transfer is complete and a signal appears via the No. 6 circled command line to the reload input terminal 206, the twodecade counter 200 is reloaded with its original count which had been stored in the storage register 202.

The count inthe two-decade counter is thus preserved in a manner which allows it to be combined with a subsequent average length correction signal which may subsequently be derived inthe average length accumulator section 58.

Signals representative of successive outputs from the correction counter 170 in the average length correction section 56, are combined algebraically inthe twodecade counter200 of the average length accumulator section 58 in substantially the same manner that successive deviation signals are algebraically combined in the two-decade counter in the average length correction section 56. Thus, where a previous average length correction signal is present in the two-decade counter 200,11 subsequent correction signal is transferred from the average length correction section 56 to the two-decade counter 200 to change the count in that counter. This correction signal is applied in the form of clock pulses via the No. 3 circled command line. These signals, if representative of an average deviation, whose algebraic sign is different then that of the algebraic sum of the accumulated average deviation count in the counter, are applied via the AND-NOT gate circuit 218 and the OR gate circuit 216 to the down-count terminal 210 to count down toward zero the count in the twodec ade counter 200. Should the correction signal being applied in this manner have a magnitude which exceeds the previous count in the two-decade counter 200, a borrow signal will appear at the borrow output terminal 226 when thetwo-decade counter 200 reaches a zero count. This signal is applied to the common input terminal 224 of the direction flip-flop circuit 222 which, in turn, causes a signal to be applied to the EXCLU- SIVE OR gate circuit 214 and from there to the inhibit terminal of 220 of the AND-NOT gate circuit 218 and to the second input terminal of the AND gate circuit 212. As a result, the remaining clock pulses to be transferred to the two-decade counter 200 from the No. 3 circled command line are applied via the AND gate cir cuit 212 and the rip-count input terminal 208, thereby causing the counter200 to count upwardly again.

The presence of a signal at the borrow output terminal 226 of the two-decade counter 200 during the transfer of information from the average length correction section 56 will not advance the program, even though such signals appear on the No. S callback line. This is because the program control section 54 is arranged to advance the program only when a callback signal appears on a callback line which has the same number as'the circled command line from which signals are presently being generated.

In the event that the average length correction signal is of the same algebraic sign as that of the accumulated count in the counter 200 then a count up signal is generated in the EXCLUSIVE OR gate circuit 214 and is applied to the inhibit terminal 220 of the AND NOT gate 218 and to the remaining input of the AND gate 212. This closes the AND NOT gate and opens the AND gate so that the clock pulse signals on the No. 3 circled command line are directed to the up-count input terminal 208 of the two-decade counter 200. The count-up signals from the EXCLUSIVE OR gate circuit 214 occur wheneverthe algebraic sign of the count in the counter is the same as that of the correction count 20 When the direction flip-flop circuit 222 is in its opposite state, that is, when the signal of the accumulated count in the counter 200 is positive or long, then no signal is applied to the EXCLUSIVE OR gate circuit 214 from the direction flip-flop circuit 222 and the reverse of the above described AND and AND NOT gate control occurs for clock pulses on the No. 3 circled command line. It will be appreciated from the foregoing that the count in the counter 200 is equal in magnitude to the algebraic sum of the various signals supplied to it.

As indicated previously, both phase corrections (i.e., corrections made only once based on each preceding measured deviation) and average length corrections (i.e., corrections made continuously based on a statistical processing or several preceding deviations) are made by adjustment to the difierential transmission 26. In the event that the adjustment called for exceeds a certain predetermined limit, the system will automatically generate a signal indicating that an override adjustment should be made to the Reeves mechanism 28 so as to relieve the amount of adjustment that must be made to the differential transmission 26 between each successive operation of the knife blades 32. As shown inFIG 6B, the signals in the two-decade counter 200 are applied, as indicated by arrows 203A, to an average length correction AND gate 230. This AND gate is set to produce an output when the signals applied via the line 203 A indicate a predetermined magnitude. When this occurs, the signals are transferred from the AND gate 230 to an excessive correction input terminal 232 of an excessive correction flip-flop circuit 234. Signals applied to this terminal cause the flip-flop circuit 234 to produce a signal on an output terminal 236 which may be used many manner to indicate that the system should be adjusted by the insertion of a correction to the Reeves mechanism 28. The excessive correction flip-flop circuit 234 is also provided with a second input I terminal 238 which receives signals via a delay circuit applied via the No. 3 circled command line. Each time i the count in the two-decade counter 200 passes through zero, the resultant signal on its borrow terminal 226, which is applied to the input 224 of the direction flip-flop circuit 222, causes the flip-flop circuit to change state; so that the output from the flip-flop circuit'which is applied to the EXCLUSIVE OR gate circuit 214 occurs only when the algebraic sign of the count in the counter is negative or short. If, during this time, the sign of the correction signal from the No. 3 circled command line is also short, no long signal will appear at the other input of the EXCLUSIVE OR gate circuit so that it will generate an output to open a path from the No. 3 circled command line to the up count terminal 208 of the counter. If, on the other hand, the sign of the correction signal from the No. 3

circled command line is long, a long signal will appear at the remaining input terminal of the EXCLU- SIVE OR gate circuit and it will not then produce an output. In this case the pulse on the No. 3 circled command line will be directed to the down count input terminal 210.

240 from the borrow output terminal 176 of the correction counter 170 in the average length correction section 56. Whenever the count present in the correction counter 170 is transferred to the two-decade counter .200 in the average length accumulator section 58, the

switched to produce an output signal indicating a correction should be made to the Reeves mechanism 28. Shortly thereafter, a signal will be applied to thesecond input terminal 238 of the excessive correction flip-flop circuit 234 to remove any previously generated output signal.

THE CORRECTION SECTION The purpose for the correction section 60 is to generate a total correction signal corresponding to the algebraic sum of the phase correction signal produced in the measurement section for the deviation of the irnme diately preceding cut and the average length correction signal stored in the average length accumulator section;

and to convert the total correction signal into a corre sponding rotation of the shaft for adjustment of the differential transmission 26. This is done during the in terval between each successive operation of the knife blades 32. As indicated previously, the total correction signal is a composite of a phase deviation signal and an average length deviation signal. The phase deviation signal is generated in the measurement section 52 following the occurrence of each operation of the knife blades 32. The amount of this deviation is transferred from the decade counters 72 in the measurement section to further decade counters 250 in the correction section. The average length correction signal, as indicated previously, is maintained in the average length accumulator section 58 and is updated periodically by operation of the average length correction section 56. Thus, after the occurrence of each operation of the knife blades 32 and the obtaining of a count in the measurement section 52 corresponding to the deviation of the actual knife cut location from the location of its corresponding desired line of cut, this deviation, as represented in the decade counters 72 of the measurement section, is transferred to the decade counters 250 of the correction section 60. Following this, the count which has been maintained in the average length accumulator section 58 is combined algebraically with the phase deviation signal in the decade counters 250. The resulting count is then used to produce a corresponding rotation of the shaft 70 for adjustment of the differential tranmission 26.

The decade counters 250 in the correction section 60 are provided with an up-count input terminal 252 and a down-count input terminal 254. The up-count input terminal is connected to receive clock pulses via a first OR gate circuit 256 from the No. l circled command line. Signals are also applied via the first OR gate circuit 256 from an AND gate circuit 258. One input terminal of the AND gate circuit 258 is connected to receive signals from an EXCLUSIVE OR gate circuit 260. One input terminal of this EXCLUSIVE OR gate circuit 260 is connected to receive signals from the direction flip-flop circuit 222 in the average length accumulator section 58 whenever that direction flip-flop circuit applies signals to its own EXCLUSIVE OR gate circuit 214. The other input terminal of the EXCLU- SIVE OR gate circuit 260 is connected to receive signals from a long correction output terminal 262 of a direction flip-flop circuit 264. i

The down-count input terminal 254 of the decade counters 250 is connected to receive signals via a second OR gate circuit 266 from either an AND-NOT gate circuit 268 or from a variable divider circuit 270. The AND-NOT gate circuit 268 has an inhibit terminal 272 which is connected to receive outputs from the EX- CLUSIVE OR gate circuit 260. The other input terminal of the AND-NOT gate circuit 268 and the remaining input terminal of the AND gate circuit 258 are each connected to receive clock pulse signals from the No. circled command line. The direction flip-flop circuit 264 is provided with a long measurement input terterminal 278. The borrow terminal 278 of the decade counters 250 is also connected, as indicatd, to the No. 7 callback line.

The variable divider 270 operates to produce a series of pulses at an output terminal 280 at a rate corresponding to some predetermined sub-multiple of pulses applied to an input terminal 282 thereon. The pulses which are applied to the input terminal 282 are generated by an oscillator 284 and these signals pass through an AND gate 286 to the variable divider 270. The AND gate circuit 286 is also connected to receive signals from the No. 7 circled command line.

The output of the oscillator 284 which passes through the AND gate 286 and is applied to the variable divider 270 is also applied to a divider circuit 288 which produces a series of pulses at a predetermined sub-multiple of the applied oscillator frequency. These pulses are applied to one input terminal each of an AND gate circuit 290 and an AND-NOT gate circuit 292. The AND-NOT gate circuit 292 has an inhibit terminal 294 which, together with the remaining input terminal of the AND gate circuit 290, is connected (in a manner not shown) to receive long correction signals from the long output terminal 110 of the direction flip-flop circuit 98 in the measurement section 52. The

output of the AND gate circuit is applied to a CW (clockwise) input terminal 296 of an electrohydraulic stepper motor system 298. The output signals from the AND-NOT gate circuit 292 are applied to a CCW (counter-clockwise) input terminal 300 of the stepper motor system 298. The steppermotor system operates to produce rotation of the shaft 70 in a direction corresponding to the particular one of its input terminals 296 and 300 upon which signals appear and in an amount corresponding to the number of pulses received at that terminal.

The correction section 60 operates in the following manner. When a out has been made on the web l0, the

, deviation of the location of that cut from a desired cut the counters 72 of the measurement section to the deminal 274 which is connected in a manner not shown to receive signals corresponding to the occurrence of long measurements from the long output terminal 110 of the direction flip-flop circuit 98 in the measurement section 52. The direction flip-flop circuit 264 is also provided with a common input terminal 276 which is connected to a borrow terminal 278 on the decade counters 250. The direction flip-flop circuit 264 changes state upon each reception of a signal at its common input terminal 276; and this occurs whenever the decade counters 278 pass through a zero count condition and thereby produce a signal at their borrow cade counters 250 of the correction section 60. This transfer occurs during that portion of the program when clock pulse signals appear on the No. l circled command line and this transfer becomes completed when the count on the decade counter is reduced to zero and it produces a borrow signal at its borrow terminal.

The transfer of the deviation count in the measurement section 52 to the decade counters 250 of the correction section constitutes the generation of a phase trend correction signal which is present in the average length accumulator section 58. This algebraic combilength accumulator section 58 to the correction section The transfer from the average length accumulator section takes place during that portion of the program when clock pulses are being generated on the No. 5 circled command line. These pulses are applied simultaneously to the down count terminal 210 of the counters nation takes place simultaneously with the transfer of a count from the decade counters 200 of the average 200 in the average length accumulator section 58 and to one terminal of the AND gate circuit 258 and of the AND NOT gate cirucit of the correction section 60. These clock pulses continue until the count in the average length accumulator section has been reduced to zero. This produces a corresponding count in the correction section 60. The pulses applied to the correction section 60 are directed to either the up count terminal 252 or to the down count terminal 254 of the counter 250 depending upon which of the

gates

258 or 268 is opened. This is controlled by the output of the direc tion flip-flop circuit 222 in the average length accumulator section. As shown in FIG. 63, this circuit is connected to supply a signal to the EXCLUSIVE OR circuit 260 in the correction section 60. This circuit will receive a short signal from the average length accumulator section whenever the count stored in that section indicates a long average deviation correction. Now if the phase deviation happens to be long," the direction flip-flop 264 in the correction signal will be energized at its terminal 274 so that it will produce a long signal at its output terminal 262 and at the EXCLU- SIVE OR gate 260. In such a case the

gates

258 and 260 will be switched so that the transfer of the average length correction signal to the counter 250, which is applied via the No. circled command line will be dirooted to the down count terminal 254. It will be seen therefore that the long phase correction signal count and the short average length correction signal count are differenced in this case. Whenever the phase and average deviation correction signals are both long or both short the

gates

258 and 268 are switched to di rect the average length correction signals applied via the No. 5 circled command line to the up count input terminal 252 of the counters 250. When the phase and average length correction signals are different, i.e., when one is short while the other is long the average length correction signals applied via the No. 5 circled command line will be directed to the down count terminal 254 of the decade counters 250. In this latter case, the count in the counters 250 may be reduced by zero. At the zero count a signal will appear on the borrow output terminal 278 and will be applied to switch the correction direction flip-flop circuit 264 to switch the

gates

258 and 268 so that the remaining signals will be applied to the up count terminal of the counters 250.

When the phase and average deviation correction signals have been thus combined a signal appears on the No. 7 circled command line to the AND gate 286. This opens the gate and allows the pulses from the oscillator 284 to be applied, via the divider circuit and the AND and AND-NOT gate circuits 290 and 292 to ei ther the CW or CCW terminals 296 or 300 of the electro-hydraulic stepper motor system 298 to turn the shaft 70 by an amount corresponding to the number of pulses applied to the system 298 while the gate 286 is opened and ma direction corresponding to which of the gates 290 or 292 is opened. This direction is controlled by the appearance or absence of a signal on a same, it will be obvious to those skilled in the art, after As pulses from the oscillator 284 are applied via the AND gate 286 to the electro-hydraulic stepper motor system 298, they are also applied via the variable divider 270 to the down count terminal 254 of the counters 250. These pulse signals continue to be applied to this counter until its count is reduced to zero whereupon a signal appears at its borrow terminal 278 and on the No. 7 callback line. This advances the program and removes the signal from the No. 7 circled command line to close the gate 286 and stop further application of oscillator pulses to the electrohydraulic stepper motor system 298 and to the counters 250.

It will be appreciated that the shaft 70 of the differential transmission mechanism is adjusted by an amount and in a direction corresponding to the combination of phase and average deviation correction signals supplied to the counters 250 in the correction 60 following each knife cut.

It may be desired to control the electro-hydraulic stepper motor system 298 so that instead of abruptly stopping operation when the gate 286 is closed, it will begin to slow down and come to a more gradual stop at this point. For this purpose there is provided a comparator 306 which receives a continuous sampling of the count present in the counters 250. The comparator 306 is hard wired to establish a predetermined slowdown count. The count inthe counters 250 is compared with this slowdown count and when the count in the counters 250 falls below this number, the comparator 306 sends a near zero signal to the oscillator 284 to reduce its output frequency.

Having described the invention with particularity with reference to the preferred embodiment of the understanding the invention, that various changesand modifications may be made therein without departing from the spirit and scope of the invention; and the appended claims are intended to cover such changes and modifications as are within the scope of the invention.

long correction line 304 which signal is applied to one terminal of the AND gate circuit 298 and to the inhibit terminal 294 of the AND-NOT gate circuit 292. The

long correction line 304 is connected (in a manner not shown) to the long correction terminal 262 of the correction direction flip-flop 264.

What is claimed is:

l. A method for controlling a device which operates eration of the device based upon a first combination of the deviation signal generated during the immediately preceding interval and a first trend signal, maintaining said first trend signal for a succession intervals, while comparing only the deviation signals generated during said succession to produce a new trend signal, adjusting each of a further succession of intervals of operation following the production of the new trend signal, the last mentioned adjusting being based upon a new combination of signals including the deviation signal generated during the immediately preceding interval and the new trend signal and maintaining the new trend signal while comparing the deviation signals developed during said further succession of intervals.

2. A method according to claim 1, wherein said adjusting comprises the steps of generating first correction signals corresponding to said deviation signals in' each interval, generating second correction signals corresponding to said trend signal being maintained, algebraically combining said first and second correction signals in each interval to produce a total correction signal and adjusting the interval of operation of said machine according to said total correction signal.

3. A method according to claim 2, wherein each said second correction signal is stored and is combined with the first correction signal produced in each of the intervals following production of said second correction signal.

4. A method according to claim 1 wherein said new combination of signals further includes said first trend signal algebraically combined with said new trend signal.

5. Apparatus for regulating a system which operates on a medium in a sequence of intervals so that the intervals of device operation coincide with predetermined intervals on said medium, said apparatus comprising means for generating deviation signals corresponding to the deviation of each operation interval from an associated predetermined medium interval, phase correction means operative to produce a phase correction signal corresponding solely to the immediately preceding deviation signal, average deviation correction means operative to produce an average deviation correction signal in response to the algebraic sum of successive statistical combinations of separate groups of successive deviation signals and means for adjusting each interval of operation of said system by an amount corresponding to a combination of phase correction signals and said average deviation correction signals.

6. Apparatus according to claim 5, wherein said means for adjusting each interval of operation of said system includes correction signal combining means connected to receive said phase correction signal and said average deviation correction signal and to combine said correction signals to produce a total correction signal and means responsive to said total correction signal for producing a corresponding adjustment in the interval of operation of said system.

7. Apparatus according to claim 6, wherein said means for generating said correction signals includes means for generating series of pulses corresponding in number to the magnitude of deviation and means for indicating the direction of deviation and wherein said correction signal combining means comprises pulse counter means connected to receive said pulses and to combine said pulses algebraically.

8. Apparatus according to claim 5, wherein said means operative to produce a phase correction signal includes correction signal storage means operative to retain an average deviation correction signal following 55 combination of such signal with a phase correction signal.

9. Apparatus according to claim 5, wherein said means operative to produce an average deviation correction signal comprises first and second signal combining means, each operative to combine signals applied thereto in an algebraic manner, in reg ard to magnitude and sign, with signals previously applied to said being connected to receive and combine algebraically each first signal combining means output.

10. Apparatus according to claim 9, wherein said first signal combining means includes signal accumulator 'means and means operative to clear said signal accumulator means upon the occurrence of each first signal combining means output.

11. A method of controlling a process which is carried out in successive operations, said method comprising the steps of producing, after each operation of the process, a deviation signal representative of the magnitude and direction of the deviation of the result of the operation from a desired result, cumulatively combining each successive deviation signal in an algebraic manner with previously produced deviation signals to produce a new cumulative result following each operation, comparing each cumulative result with a different norm corresponding to the number of deviation signals which have been combined, and adjusting said process whenever any cumulative result exceeds its associated tervals, said apparatus comprising means for producing deviation signals corresponding tothe deviation of the result of each operation from a desired result, combining means for algebrically combining each deviation signal as it occurs with the algebraic total of combined preceding deviation signals to produce a new total upon the occurrence of each deviation signal, system operation counting means for producing different outputs corresponding to successive operations of said system, plural gate circuits connected to be opened by the different outputs of said counting means, said gate circuits each being further connected to respond, when opened, to a different count level in said combining means and means responsive to the occurrence of a response from any of said gate circuits for producing a correction in said system.

15. Apparatus according to claim 14, wherein the last-mentioned means includes means for clearing said signal combining means, said first signal combining means being connected to receive and combine algebraically the phase deviation signals of seperate groups and to produce a corresponding first signal combining means output, said second signal combining means combining means and for clearing said system operation counting means upon the occurrence of a response from any of said gate circuits.

16. Apparatus according to claim 14, wherein the last-mentioned means includes means for generating different correction signals corresponding to the particular gate circuit which responds to the count level in said combining means.

17. A method for controlling a web cut system wherein the location of each web cut is adjustable by an amount corresponding to the degree of rotation, which is applied between successive web cuts, to the input of a differential mechanism connected along a drive which synchronizes longitudinal web movement with operation of a cutter through which the web is moved, said method comprising the steps of producing deviation indications representative of the deviation of the actual location of each web cut from a desired cut location, deriving a first average deviation correction signal by statistically processing a series of deviation indications and preserving a phase correction signal corresponding to each immediately preceding deviation indication, during each interval between successive operations of said cutter following the deviation of said average deviation correction signal, rotating the input of said differential mechanism by an amount corresponding to a composite of said average deviation correction signal and the phase correction signal derived from the immediately preceding deviation indication, statistically processing a series of subsequent deviation signals which occur following the derivation of said first average deviation correction signal and to produce a subsequent average deviation correction signal and modifying said first average deviation correction signal by an amount corresponding to said subsequent average deviation correction signal.

18. A method according to claim 17, wherein said composite of said average deviation correction signal and said phase correction signal is derived by storing said average deviation correction signal and algebraically combining each phase correction signal with a signal equal to the stored average deviation correction signal during the interval immediately following the production of said phase correction signal.

19. A method for controlling a system which operates atsuccessive intervals to produce a series of successive results, said method comprising the steps of measuring each result so-produced and comparing same with a predetermined desired result to obtain a separate deviation indication for each operation, statistically processing a group of said deviations to produce a first averagc deviation indication, producing a first average deviation correction in said system, corresponding to said first average deviation indication, for subsequent successive operations of said system, statistically processing a further group of said deviations which occur under the influence of said first average deviation correction to produce a second average deviation indication and producing a second average deviation correction in said system corresponding to said second average deviation indication for successive operations of said system subsequent to the production of said second average deviation indication.

20. A method according to claim 19, wherein said second average deviation correction corresponds to the algebraic sum of said first and second average deviation indications.

21. Apparatus for regulating the operation of a system which operates at successive intervals to produce a series of successive results, said apparatus comprising deviation signal producing means operative to produce a deviation signal following each operation of the system, said deviation signal corresponding to the deviation of the actual results of the operation from a predetermined desired result, signal combining means operative to combine several successive ones of said deviation signals and toproduce a correction signal corresponding to a statistical average of said deviation signals, correction signal storage means for maintaining a tem, means for clearing said signal combining means in response to the production of a correction signal thereby and means for applying subsequently produced deviation signals to the signal combining means after clearance thereof and during each said subsequent interval of operation of said system.

Claims

1. A method for controlliNg a device which operates on a medium in a sequence of intervals so that the intervals of device operation coincide with predetermined intervals on said medium, said method comprising the steps of detecting each interval on said medium, detecting each operation of said device, generating deviation signals based upon each detection corresponding to the deviation from coincidence of device operation and medium interval, adjusting each interval of operation of the device based upon a first combination of the deviation signal generated during the immediately preceding interval and a first trend signal, maintaining said first trend signal for a succession of intervals, while comparing only the deviation signals generated during said succession to produce a new trend signal, adjusting each of a further succession of intervals of operation following the production of the new trend signal, the last mentioned adjusting being based upon a new combination of signals including the deviation signal generated during the immediately preceding interval and the new trend signal and maintaining the new trend signal while comparing the deviation signals developed during said further succession of intervals.

2. A method according to claim 1, wherein said adjusting comprises the steps of generating first correction signals corresponding to said deviation signals in each interval, generating second correction signals corresponding to said trend signal being maintained, algebraically combining said first and second correction signals in each interval to produce a total correction signal and adjusting the interval of operation of said machine according to said total correction signal.

8. Apparatus according to claim 5, wherein said means operative to produce a phase correction signal includes correction signal storage means operative to retain an average deviatioN correction signal following combination of such signal with a phase correction signal.

9. Apparatus according to claim 5, wherein said means operative to produce an average deviation correction signal comprises first and second signal combining means, each operative to combine signals applied thereto in an algebraic manner, in regard to magnitude and sign, with signals previously applied to said signal combining means, said first signal combining means being connected to receive and combine algebraically the phase deviation signals of seperate groups and to produce a corresponding first signal combining means output, said second signal combining means being connected to receive and combine algebraically each first signal combining means output.

10. Apparatus according to claim 9, wherein said first signal combining means includes signal accumulator means and means operative to clear said signal accumulator means upon the occurrence of each first signal combining means output.

11. A method of controlling a process which is carried out in successive operations, said method comprising the steps of producing, after each operation of the process, a deviation signal representative of the magnitude and direction of the deviation of the result of the operation from a desired result, cumulatively combining each successive deviation signal in an algebraic manner with previously produced deviation signals to produce a new cumulative result following each operation, comparing each cumulative result with a different norm corresponding to the number of deviation signals which have been combined, and adjusting said process whenever any cumulative result exceeds its associated norm.

12. A method according to claim 11 further including the step of beginning a new cumulative combining of successive deviation signals whenever any cumulative result exceeds its associated norm.

13. A method according to claim 11, wherein said process is adjusted by an amount corresponding to the number of deviation signals which have been combined to produce the adjustment.

14. Apparatus for use in regulating a system which performs a series of similar operations in successive intervals, said apparatus comprising means for producing deviation signals corresponding to the deviation of the result of each operation from a desired result, combining means for algebrically combining each deviation signal as it occurs with the algebraic total of combined preceding deviation signals to produce a new total upon the occurrence of each deviation signal, system operation counting means for producing different outputs corresponding to successive operations of said system, plural gate circuits connected to be opened by the different outputs of said counting means, said gate circuits each being further connected to respond, when opened, to a different count level in said combining means and means responsive to the occurrence of a response from any of said gate circuits for producing a correction in said system.

15. Apparatus according to claim 14, wherein the last-mentioned means includes means for clearing said combining means and for clearing said system operation counting means upon the occurrence of a response from any of said gate circuits.

19. A method for controlling a system which operates at successive intervals to produce a series of successive results, said method comprising the steps of measuring each result so-produced and comparing same with a predetermined desired result to obtain a separate deviation indication for each operation, statistically processing a group of said deviations to produce a first average deviation indication, producing a first average deviation correction in said system, corresponding to said first average deviation indication, for subsequent successive operations of said system, statistically processing a further group of said deviations which occur under the influence of said first average deviation correction to produce a second average deviation indication and producing a second average deviation correction in said system corresponding to said second average deviation indication for successive operations of said system subsequent to the production of said second average deviation indication.

21. Apparatus for regulating the operation of a system which operates at successive intervals to produce a series of successive results, said apparatus comprising deviation signal producing means operative to produce a deviation signal following each operation of the system, said deviation signal corresponding to the deviation of the actual results of the operation from a predetermined desired result, signal combining means operative to combine several successive ones of said deviation signals and to produce a correction signal corresponding to a statistical average of said deviation signals, correction signal storage means for maintaining a correction signal produced by said signal combining means, means for successively applying the correction signal from said correction signal storage means to adjust each subsequent interval of operation of said system, means for clearing said signal combining means in response to the production of a correction signal thereby and means for applying subsequently produced deviation signals to the signal combining means after clearance thereof and during each said subsequent interval of operation of said system.