US2873117A

US2873117A - Register control of moving webs

Info

Publication number: US2873117A
Application number: US575277A
Authority: US
Inventors: Crosfield John Fothergill; Gold Daniel
Original assignee: CROSFIELD J F Ltd
Current assignee: CROSFIELD J F Ltd; J F CROSFIELD Ltd
Priority date: 1955-04-01
Filing date: 1956-03-30
Publication date: 1959-02-10
Anticipated expiration: 1976-02-10

Description

Feb. 10, 1959 I J. F. CROSFIELD ET AL REGISTER CONTROL OF MOVING WEBS 8 Sheets-Sheet 1 Filed March so, 1956 A Home y I n venton:

y Mt;

Vmm

Feb. 10, 1959 J. F. CROSFIELD ETAL 2,873,117

REGISTER CONTROL'OF MOVING WEBS Filed March 30, 1956 8 Sheets-Sheet 2 Inventors AAAAAI DON 3w NR 0% www www I: Now 8m QM mm wmm 0% com I mam mam 0km SN N @C NON Feb. 10, 1959 J. F. CROSFIELD ET AL 2,873,117 REGISTER CONTROL OF MOVING WEB-S Filed March 30, 1956 8 Sheets-Sheet 3 FIGS.

A ltorney REGISTER CONTROL OF MOVING WEBS Filed March 30, 1956 8 Sheets-Sheet 4 1' nventor; F

M 70% 0am Attorney Feb. 10, 1959 Filed March 30, 1956 8 Sheets-Sheet 5 Inventors- MM fiq; a. mad- Attorney Feb. 10,1959 J. F. CROSFIELD ETAL I ,8

REGISTER CONTROL OF MOVING WEBS 8 Sheets-Sheet 6 Filed March 30, 1956 A." A" m man i w v w 30w l I, Av m m wmh M mp2. F NM vmw I Q L QM mum Vhm DVD-A.) 00? 3 w? 1nventom By 6M W Attorney Feb. 10, 1959 J. F. CROElFlELD ET AL REGISTER CONTROL OF MOVING WEBS 8 Sheets-Sheet 7 Filed March 30, 1956 I J. F. CROSFIELD ET AL 2,873,117

REGISTER CONTROL. OF MOVING WEBS Feb. 10, 1959 8 Sheets-Sheet. 8

Filed March 30, 1956 OON 50mm A Horn 2 y United St tes Patent fiice 2,873,117 Patented Feb. is, 1959 REGISTER CONTROL OF MOVING WEBS John Fothergill Crosfield and Daniel Gold, London, England, assignors to J. F. Crosficld Limited, London, England, a British company Application March 30, 1956, Serial No. 575,277

"Claims priority, application Great Britain April 1, 1955 17 Claims. (Cl. 2712.6)

This invention relates to systems of register control for moving webs. Such control is required in machines for performing various operations upon continuous flexible webs, for example, the printing of paper webs as in multicolour printing, wherein impressions in inks of different colours have to be superimposed in accurate register, or again in cutting or embossing webs at places which have to be in a predetermined accurate relationship to other impressions or operations. Another use of such systems lies in the printing of a web taken from storage on a bobbin or storage reel, with impressions in precise register with those already printed on the same web. Another problem which is sometimes more diflicult to deal with occurs in feeding a preprinted web either alone or with others immediately the latter are printed into another machine such as, for example, a folding machine, wherein the web has to be maintained in accurate register with the cyclic operation of the machine, which usually involves the feeding of an exactly determined length of the web at each cycle of operation of the folding or other machine.

One method employed for maintaining and correcting register of moving webs involves the use of register marks upon the web at positions definitely related to the positions of two or more areas printed successively, so that the distance apart of two marks represents the amount of error in register of two printing operations, and the operation of a register-correcting mechanism to effect a correction dependent upon the distance apart of the two marks. Such a system is disclosed in the specification of British Patent No. 641,830 and a somewhat modified system is described in patent application Serial Number 465,684, now U. S. Patent No. 2,802,666. As described in those cases, the marks on the web pass a scanning head containing two photoelectric cells and a measure of the error is obtained by electronic valve circuits in dependence upon the timing of the arrival from the photoelectric cells of impulses corresponding to the two marks. In those cases, the correction is made by an electric correction motor which is caused to run for a period which is substantially proportional to the amount of error at any time and this effects small corrections which may be positive or negative as required. These corrections may be applied to compensating jockey rollers or on the printing cylinders. However, when a correction is applied, owing to the elasticity of the paper web, there is a small time delay before the correction takes full effect and this is allowed for by arranging that after a correction has been applied, a corresponding time is allowed to elapse before a further correction is made.

However, in the case of a preprinted web, that is to say, a web that has been printed in one or more colours and then stored and which has to be brought into step in a further operation, the problems involved are some-- what different. In some cases, where .a preprinted web has tobe subjected to. cutting, .punching or perforating orhas to be used for ag-making it is possible to, apply corrections to the cutters or other tools, or directly to the web by using draw rollers, and the corrections immediately take effect. When a preprinted web, however, has to be brought into step with the operation of, for example, a folding machine .into which other Webs which have just been printed are being fed, corrections cannot be applied to the folding or other after-treatment mechanism because corrections of the folding mechanism in respect of the preprinted web would throw the other webs out of register. Thus, the only way to apply corrections in this case is by variable stretching of the preprinted web which, as in the systems mentioned above, involves a time delay before the correction takes full effect. Moreover, unlike multicolour printing consecutively on a web, it is possible for the register to change cumulatively.

The present invention is concerned with solving the problems in that kind of control although it is also applicable to register control in multicolour printing of the kind already referred to.

When a continuous paper web is printed in a rotary press, the basic impression is repeatedly printed along the whole length of the web and the length of each impression at the instant of printing is determined only by the circumferential length of the printing cylinder. Owing to the elasticity of the paper, after printing the length of an impression may differ from that of the circumference of the printing cylinder if the tension of the web after printing differs from that at which printing took place. Furthen the action of heat used for drying the ink may permanently shrink the web after printing. Yet again, if a web is re-reeled and stored after printing, the length of each impression may be altered by humidity.

Whena preprinted web is passed into a folder together with one or more other webs which have been freshly printed, a certain fixed length of these webs is drawn through the folder by its draw rollers at each folder operation. For the preprinted web to remain in register with the operation of the folder and with the other web or webs, it is essential that at each operation exactly one impression on the web shall be drawn through the folder, for if that is not the case, the register will change by a small amount and the error is cumulative.

The length of preprinted web drawn through by the feed rollers of a foldermay depart from the standard length of one impression due to various causes. Thus, the length of paper web drawn through may differ from the circumferential length of the preprinting" cylinder; the tension under which the preprinted web is drawn through a folder may be different from that under which the preprinting is effected or again, the length of an impression may be affected by heat or humidity. However, so long as the length of a preprinted impression with the web unstretched is less than the length of webs drawn through at each operation of the folder, it can be ensured that a length exactly equal to that of one impression is drawn through at each operation @by stretching the preprinted web as it goes through the draw rollers of the folder, and this may be done by applying a suitable tension to the preprinted web at the draw rollers to stretch it sufficiently.

if the preprinted web is fed directly from a reel to the draw rollers of a folder, one method of adjusting its tension is to apply a brake to the reel and to adjust the braking action. This, however, is not always convenient, inwhichcaseflhe web from a reel may be passed through a pair of control rollers on its way to the ,draw rollers an infinitely variable speed gearing to which the register error correction is applied, as will be described later.

If there is an error in register between the operation of the folder and the preprinted web, and the speed of the control rollers is set to the value which gives the correct tension to provide the correct length of web per revolution of the folder, clearly the error will persist at a constant value. To correct such a register error, two factors must be taken into account, namely, the value of the constant error and the fact that when corrections are applied to the speed of the control draw rollers, the tension of the web does not change immediately, owing to its elasticity. For that reason, in all such cases, the corrections are made periodically to allow time for each to take partial effect before the next correction is made. It has been found that in systems of the kind which have been discussed the register control is improved if in computing the correction to be applied the derivative factor is taken into account, that is to say, if the correction factor is of the form includes a term representing the instantaneous rate of change of the error. In electrical control systems utilising derivative control it is usually possible to obtain an electric signal representing the derivative factor by applying a signal representing the quantity to be controlled to a differentiating circuit. In register control systems, however, difiiculties are encountered if an attempt is made to measure the derivative factor in this manner. These difficulties are caused principally by the fact that the rate of change of the error is very small, and that the error indications are in the form of sampled data and are not sufficiently continuous. It is thought that it might be possible to obtain a derivative signal by applying the error signal to a differentiating circuit if the error indications followed each other at sufliciently short intervals. This, however, would involve the printing of a large number of register marks in each preprinted unit of a preprinted web, and is therefore not desirable.

These difficulties are overcome in the present invention by using a control factor obtained by measuring the error at intervals, which may correspond to the passage of several preprinted units, and subtracting to find the change in the error between two measurements.

According to the present invention, therefore, in apparatus for the automatic register control of continuous moving webs, a register error is reduced by periodically adjusting a register correction means in accordance with both the magnitude of the error and the change in the error during a short period preceding an adjustment of the register correction means. This may be expressed by stating the control factor as (EK.AE), where E represents the magnitude of the error. The magnitude of the error may be determined by means of photo-electric devices and electronic circuits as in the prior British patent specification and the patent application mentioned above and the change of the error between two measurements may be determined by further electronic valve circuits. As already indicated, the correction is made periodically in order to allow each correction to take partial effect before making the next adjustment. With the arrangement according to the invention, only a single register mark need be printed for each preprinted unit on a preprinted web.

In the case of feeding a preprinted web-to a folder or like mechanism, the control may be efiected on an electric motor which is connected to change the ratio of an colour printing, an electric motor similarly controlled applies the correction to the web through a compensating pulley or through the printing cylinder. By a suitable choice of the constants of the system, the control can be made substantially to reduce the error itselfandl the rate of change of error to zero.-

As an example of the operation of apparatus according to the invention, if there is an error which is chang-- ing and the control factor (E-K.AE) is zero, no correction will be introduced but at the next correction time, the error will have changed so that the control factor is. no longer zero and a correction will be made. After a few corrections, both the error and rate of change of error are reduced to negligibly small amounts.

A convenient method of applying the control factor to the correcting electric motor is to provide a pair of condensers, one of which is charged to a potential representing the error proportional term and the other to a potential representing the change in the error. In applying these potentials to thecircuit of the correction motor, it is preferred to connect the two condensers in parallel rather than in series, since by using condensers of different capacities, the potentials adjust themselves and thus the constant K may be readily introduced into the control factor.

In order that the invention may be more fully understood, an example of apparatus according to the invention for the automatic register control of a preprinted web which is being fed to a folding machine will now be described with reference to the accompanying drawings. Parts of the circuit shown in the accompanying drawings have been fully described in the specification of our copending pzftent application No. 465,684 (now Patent No. 2,802,666), and these will be described only briefly in the present specification.

In the drawings:

Figure 1 shows diagrammatically the arrangement of the apparatus;

Figure 2 is a block diagram of the control circuits;

Figure 3 is a circuit diagram of the scanning head;

Figure 4 shows a magnetic switch for generating the master register and gating pulses;

40 Figure 5 is a circuit diagram of an amplifying and gating circuit;

Figure 6 shows details of a circuit for producing a gating waveform;

Figure 7 shows a pulse-generating and error-integrating circuit;

Figure 8 is a circuit diagram of apparatus for measuring the rate of change of the error;

Figure 9 is a circuit diagram of the apparatus for controlling the speed and direction of rotation of the correction motor, and

Figure 10 is a circuit diagram of timing apparatus.

In Figure 1 a reel of preprinted paper 240 is mounted in a reelstand 242, and a web of paper 243 is led from the reel over a tension roller 244 and through control draw rollers 246 and further draw rollers 248 to the rollers 249 of a folding machine. Further webs 250 and 252 which have just been printed are drawn through the common rollers 248 to the folding machine. The control draw rollers 246 are driven through infinitely variable gearing 254 by a motor 256, the ratio of the gearing 254 being adjustable by means of a small motor 258. The motor 256 also drives a belt 260 acting on the periphery of the reel 240, but the belt 260 is driven at a speed slightly less than that of the web, so that it acts as a brake.

Marks on the web are scanned by a photo-electric scanning head 262, the pulses from the scanning head being applied to a control unit 264, which also receives pulses from a magnetic switch 266 which is driven in synchronisrn with the rollers 249 of the folder mechanism. A tachogenerator 268 provides the control unit 264 with a signal representing the operational speed of the folder. When the printed units on the paper web are in correct register for the operation of the folder, the impulses from the scanning head 262 and the magnetic switch 266 are in synchronism. When there is an error in the register, there is an interval between the pulses from the scanning head and the magnetic switch, and the control unit 264 generates an electric correction signal which is applied by way of the conductor 26? to the motor 258, which makes a corresponding adjustment to the ratio of the gearing 254. As a result, the speed of the control draw rollers is modified and the tension of the web between the latter rollers and the folder draw rollers is adjusted. V i hen the next correction signal is applied to the motor 258, a further adjustment, which may increase or decrease the eifect of the preceding adjustment, is made to the speed of the control draw rollers and the tension of the web. The amplitude of the correction signal is controlled in accordance with both the magnitude of the error and the change of the error between successive corrections.

The electronic circuits for generating the signal which controls the motor 258 will first be described with reference to the block diagram of Figure 2, and this will be followed by a detailed description.

The scanning head 262 (Figures 1 and 2) provides an output which includessignals corresponding to the printed matter as well as the signals corresponding to the register marks, which are printed in a small clear space which is left in the printed matter. The output of the scanning head is applied to an amplifying and gating circuit 270 by means of which the signals corresponding to the printed matter are removed, leaving only the amplified register signals.

From the magnetic switch 266, an output signal consisting of a train of master register pulses, one for each operation of the folder, is applied to an amplifying circuit 272. The magnetic switch 266 also provides a signal consisting of a train of pairs of pulses, the pulses of each pair corresponding to a portion of the clear space which contains the register mark, and this signal is applied to a pulse-forming circuit 274. This latter circuit generates a gating signal which is applied to the gating and amplifying circuit 270 to remove the signal corresponding to the printed matter.

The signal from the circuit 270, consisting of a train of pulses corresponding to the register marks on the web, and the signal from the circuit 272, consisting of a train of master register pulses, are applied to a pulse-generating circuit 276 which produces, for each revolution of the folder, a square pulse of which the pulse width is equal to the interval between the register signals from the scanning head .andthe magnetic switch. This square pulse is applied either to

conductors

278 and 280, or to

conductors

282 and 284, depending on which of the two register signals came first, that from the scanning head or that from the magnetic switch.

The pulses on the

conductor

280 or 284 are applied to an error-integrating circuit 286, which provides, either on conductor 288 or on a conductor 290, a D. C. signal the amplitude of which represents the magnitude of the error in register.

The pulses on the

conductor

278 or 282 are applied to a change-of-error circuit 292. This latter circuit, under the control of a timing circuit 294 which is actuated by pulses from the amplifying circuits 270 and'272, measures the change of error in the period between successive samplings of the register error and provides on conductors 298 and 300 a signal representing the change of error during this period. The error maybe sampled, for example, once in every five revolutions of the folder.

The error signals and the signals representing change of error, are applied to a motr-0perating circuit 302 which drives the motor 258 in such a manner as to bring the register error back to zero. v

The circuits represented by the blocks in Figure 2 are shown in detail in Figures 3-10, which will now be described. Reference will bemade to our co-pen'ding application Serial Number 465,684 (Patent No.'2,802,"666) for a fuller description of some of thesecircuits, and where components shown "in Figures 342 are similar to those shown in our said co-pending application and play a similar part in the operation ,of the apparatus, they will be given the reference numerals used in the earlier application.

The scanning head 262 (Figure 2) for which the circuit diagram is shown in Figure 3, is described in detail in our said co-pending application. Briefly, it includes a photo-electric cell 6 from which signals representing the passage of register marks and blocks of printed matter on the web are applied to the control grid of a triode 8 arranged as a cathode follower. The output of this cathode follower stage is applied by way of a conductor 12 to amplifying and gating circuits which will be described later.

The magnetic switch 266 (Figure 2) is shown diagrammatically in Figure 4. Like the switch shown in Figure 6 of the said co-pending application, it includes a shaft 52 on which is mounted a steel disc 56 carrying two radial pole pieces 57. In the present case the shaft 52 is driven in step with the operation of the folder mechanism. At each revolution of the shaft 52 the pole pieces 57 pass in sequence in front of stator pole pieces 53, the outer ends of which are connected by a permanent magnet 54. Thus, during each revolution of the shaft 52, the magnetic circuit including the pole pieces 53 and the magnet 54 is completed by each of the rotor pole pieces 57 in turn, and two pulses are generated in a coil 55 wound on one of the pole pieces 53 and are applied to an output conductor 304. These pulses are used for gating purposes, to remove from the output of the scanning head the signals corresponding to the printed matter. The relative angular position of the rotor and stator pole pieces for a given condition of the folder is such that the two pulses occur respectively just before and just after the register mark is scanned.

The magnetic switch shown in Figure 4 includes a further pole piece 306 mounted on the steel disc 56 and two further stator pole pieces 308 on one of which a coil 310 is wound. The

pole pieces

306 and 308 are arranged nearer to the axis of the shaft 52 than the pole pieces 57 and 53, so that the rotor pole piece 306 does not pass in front of the stator pole pieces 53 and the rotor pole pieces 57 do not pass in front of the stator pole pieces 308. The pole piece 306 is diametrically opposite the gap between the pole pieces 57, and during each revolution of the shaft 52, a single pulse is generated by the coil 310 and is applied to the output conductor 312, this pulse occurring in the interval between the two pulses which are generated by the coil 55.

The signals on the output conductor 12 of the scanning head and the output conductor 312 of the magnetic switch are applied respectively to two similar amplifying circuits 270 and 272 (Figure 2), the amplifier 270 being shown in Figure 5. This amplifier is similar to that shown in Figure 4 of our said co-pending application and as it is fully described in the specification of the latter, only a brief description will be given in the present specification.

The signal on the conductor 12, which includes signals representing the printed matter and pulses representing the register marks, is applied to the control grid of a variable-mu pentode 15, the output of which is applied through a triode amplifier 19 to the control grid of a gating triode 23. The anode conductor 29 of this triode is supplied with H. T. only during the interval between the two pulses from the magnetic switch which represent the portion of the clear space on the web containing the register mark, and the triode can therefore be rendered conducting only during this period. Only the pulses corresponding to theregister marks appear at the cathode of the triode, and these are applied to the control grid of an output triode 33. The anode of this triode is coupledto an A. V. C. delay and pulse-lengthening circuit whichincludes a triode .40, a diode 41 and a l capacitor 42 The diode 41 conducts in response to the register pulses and charges the capacitor 42 negatively. The discharge of the capacitor 42 through the resistor 46 is very slow since the latter has a high value. The A. V. C. is applied by way of a conductor 47 to the control grid of the variable-mu pentode 15, and in this way the amplitude of the pulses from triode 33 corresponding to the register marks is maintained substantially constant.

The triode 33 has a split anode load consisting of the

resistors

314 and 316. An output conductor 318 is connected to the junction of these resistors and receives the amplified negative pulses representing the register marks on the webs.

The amplifier 272 to which the master register pulses from the magnetic switch are applied, is similar to the amplifier shown in Figure with the exception that the conductor 29 is connected directly to the positive H. T. conductor. This is because there is no necessity for gating the signals from the magnetic switch.

The gating pulses from the magnetic switch are applied by way of conductor 304 to the circuit shown in Figure 6 which is similar to the circuit shown in Figure 6 of our said co-pending application and has been fully described in the specification of the latter. These pulses are applied through two triode amplifiers 59 and 61 to the two control grids of a double triode 63 which is arranged as a trigger circuit. The right-hand triode is normally conducting but the arrival of a negative pulse at its control grid causes the condition of the trigger circuit to be reversed, so that the anode potential of the right-hand triode rises sharply. The second negative pulse renders the left-hand triode non-conducting again and the potential of the anode of the right-hand triode is brought back to its original level. Thus there is produced at this anode a square pulse the time duration of which corresponds to the portion of the clear space which contains the register mark. This square pulse is applied by way of conductor 29 to the anode of the gating valve 23 in Figure 5 and renders the latter conducting for its duration.

The negative pulses on the conductor 318 (Figure 5) and those on an output conductor 318a of a similar amplifier through which pass the master register pulses from the magnetic switch, are applied to the circuits shown in Figure 7, the upper half of which shows the slave channel to which the conductor 318 is connected, the lower half being the master channel to which the conductor 318a is connected. The arrangement and operation of the valves 102, I14, 118 and 119 in the slave channel and the corresponding valves in the master channel are generally similar to the arrangement and operation of the corresponding valves in Figure of our said co-pending application, and will be described only briefly in the present specification. The pulses on the conductor 318 are applied through a capacitor 320 to the control grid of the left-hand valve of a double triode 102, arranged as a trigger circuit, and are also applied through a capacitor 322 to the control grid of the right-hand triode of a similarly arranged double triode 102a in the master channel. Similarly the pulses on the conductor 318a are applied through a capacitor 320a to the left-hand triode of the double triode 102a and through a capacitor 322a to the right-hand triode of the double triode 102. If the negative pulse on conductor 318 precedes that on conductor 318a it will render the left-hand triode of the double triode 102 non-conducting, thus reversing the condition of this trigger circuit, It will have no efiect on the righthand triode of the double triode 102a, since this valve is already cut off. When the master register pulse on conductor 318a arrives it will be applied through the capacitor 322a to the right-hand triode of the double triode 102 and will reset this trigger circuit. It will also be applied through the capacitor 320a to the left-hand triode of the double triode 102a, but will have no efiect on this latter trigger circuitowing to the application of a negative blocking pulse by way of the conductor to the control grid of the right-hand triode. This negative blocking pulse is derived from the cathode of a cathode follower triode 114 in the slave circuit, to the control grid of which the negative square pulses from the anode of the right-hand valve of the double triode 102 are applied.

Thus, if the pulse from the scanning head precedes the pulse from the magnetic switch, there is produced at the anode of the right-hand half of the double triode 102 a negative square pulse the duration of which corresponds to the interval between the pulses from the scanning head and the magnetic switch. No pulse is produced at the anode of the corresponding valve in the master channel. It the pulse from the magnetic switch precedes the pulse from the scanning head, a negative square pulse is pro duced at the anode of the right-hand half of the double triode 102a in the master channel, and no pulse is pro duced in the slave channel.

The output of the triode 114 is applied to an error integrating circuit which includes a diode 118, a triode 119 and a capacitor 117. The capacitor 117 is normally charged to a voltage of, for example, 28 volts, but when a negative pulse appears at the cathode of the triode 114, the diode 118 conducts and the capacitor discharges through it. In the interval between pulses the capacitor is charged through resistor 120. The voltage on the ca' pacitor 117 is applied to a triode 119 arranged as a cath' ode follower, the output signal from which is applied toa conductor 127.

Thus, if the pulse from the scanning head precedes the pulse from the magnetic switch a D. C. potential representing the error is produced on the conductor 127. If the pulse from the magnetic switch is first, a D. C. potential representing the error will be produced on the conductor 127a in the master channel.

The output of the triode 114 is also applied by way of a conductor 323, a capacitor 324 and a resistor 325 to the control grid of a triode amplifier 326, resulting in the production of positive square pulses at the anode of the latter which are applied to a conductor 327. The potential at the junction of the resistor 325 and a very high resistor 328 which is connected in series with resistor 325 between the control grid of triode 326 and the H. T. supply line, is maintained at a small positive value in order to prevent the triode from responding to any undesired signals of small amplitude appearing at the cathode of triode 114. -A similar triode 326a and conductor 327a are provided in the master channel. The pulses on the conductors 327 and 327a are used in the circuits for generating a signal representing the change of error between consecutive corrections.

Referring now to Figure 8, which shows the circuit for computing the change of error between successive corrections, the cathodes of the two halves 328 and 328a of a double triode are connected through cathode resistors to a point in a potential divider consisting of resistors 329 and 330 between a positive H. T. line 331 and an earthed conductor 332. The control grid of the triode 328 is connected through a rectifier 333 and a resistor 334 of 47,000 ohms to a terminal 335 to which a small-negative bias is applied. The control grid of triode 328a is similarly connected through a rectifier 333a and a resistor 334a to the bias terminal 335. The control grids are also connected together through two resistors 336 and 336a each of 1 megohm, the junction of these resistors being connected to a conductor 338 which receives from the tacho-generator 268 (see also Figure l) a positive potential representing the operating speed of the folder.

It will be seen that the

resistors

334 and 336 form a potentiometer between the positive conductor 338 and the terminal 335 (the rectifier 333 being arranged to present a low impedance in this direction), and the values of these resistors are such that the potential at the control conducts for the duration of the input pulse.

Two rectifiers 340 and 340a are connected between a conductor 344 and the anodes of the triodes 328 and 328a respectively, and two equal capacitors 342 and 342aare'conected between the positive H. T. conductor 331 and these two anodes, respectively. For the major portion of each operation of the folder these capacitors are uncharged, their outer plates being connected through the rectifiers 340 and 340a to a conductor 344 to which there is applied a positive potential of 200 volts. When an error pulse is received, the potential on the conductor 344 is reduced by means of a circuit which will be described later, the rectifiers 340 and 340a are therefore rendered non-conducting, and either the valve 328 or the valve 328a depending on whether the error pulse appears on conductor 327 or 327a draws current from its associated capacitor 342 or 342a for the duration of the pulse, the amplitude of the charging current being governed as a function of the speed of the folder by the potential on conductor 338. The speed of the folder must be taken into account because for a register error of a given length, the duration in time of the error pulse is inversely proportional to the speed of the web. The voltage to which the capacitor 342 or 342a is charged is a function of the product of the duration of the error pulse and the operational speed of the folder, and is therefore proportional to the length of the error, independently of the folder speed. I l

In the intervals between error pulses, the charged capacitor 342 or 342a is discharged, the potential of the conductor 344 being restored to 200 volts. The circuit which controls the potential of the conductor 344 is shown in the lower half of Figure 8.

Negative pulses from the anode of triode 33 (Figure and the anode of the corresponding triode in the master channel are applied by way of conductors 346 and 346a (Figures 5 and8) and rectifiers 348 and 348a to the control grid of a triode 350, which with a triode 352, is connected in a monostable trigger circuit. The

triodes

350 and 352 have a common cathode circuit consisting of two resistors 354 and 356 in series, and the grid of triode 352 is coupled to the anode of triode 350 through a capacitor 358. The control grid of triode 350 is connected through a resistor 360 to the junction of resistors 354 and 356, and is therefore positively biased, while the control grid of triode352 is connected to earth through a resistor 362. The triode 350 is therefore normally conducting. The arrival of a negative pulse on either conductor 346 or 346a reverses the condition of the trigger circuit, which remains in its new condition for a period determined by the time constant of the capacitor 358 and the resistor 362, after which it reverts to its normal condition. Thus the potential at the anode of triode 352 is normally 200 volts, but falls to a lower level for a period following the arrival of a negative pulse on either conductor 346 or 346a. 'The anode of triode a :pulse from the triode 33 intheslave channel or a pulse from "the correspoding triode :in the, master channel,

whichever-honours. :first. Sincerthe positive square pulse on :conductor 327 or 327a occurs simultaneously with the negative pulse on conductor ;346 orl,34 6a, ,the trigger circuit will be reversed by;thenegative;pulseat the same condition of unbalance in this double triode.

time as the leading edge of the positive square pulse causes triode 328 or triode 328a to conduct. The potential of the conductor 34 lwil1 therefore be reduced as soon as either of these triodes is rendered conducting, and as a result current will be drawn from the corresponding capacitor 342 or 342a. This charging current will continue until the end of the error pulse, and at a later moment in the cycle, the trigger circuit will reset itself and the potential of the conductor 344 will be restored to 200 volts. The capacitor 342 or 342a which has been charged by the error pulse now discharges through the corresponding rectifier 340 or 340a.

The total voltage across the two equal capacitors 342 and 342a in series is applied across the grids ofithe two halves 364 and 364a of a double triode, which are arranged as cathode followers. The output therefore appears at low impedance across the cathodes of these two triodes. A storage capacitor 366 of 0.25 microfarad is connected through a conductor 370 to the cathode of the triode 364a and through a change-over contact A5 of a relay A and a conductor 372 to the cathode of the triode 364. The relay A is operated, in a manner which will be described, when a correction is to be applied (for example, once in every five operations of the folder) and always operates while the capacitor 342 or the capacitor 342a is charged. The voltage of the capacitor 342 or 342a, representing the error length, is therefore applied to the capacitor 366, and when the contact A5 is re leased, the voltage on the storage capacitor 366 is applied across a further capacitor 368 of .005 microfarad. The upper electrode of the latter is connected through a resistor 374 to the control grid of one half 376 of a double triode, the control grid of the other half 378 being connected by a conductor 388 to the cathode of the triode 364. The triodes 37,6 and 378 are both connected as cathode followers. It will be seen that the voltage applied across the grids of these triodes at any moment is the result of substracting the voltage across the cathodes 364 and 364a (which represents the length of the error at that moment) from the voltage across capacitor 368 (which is the voltage which existed across the cathodes of triodes 364 and 364a at the time of the last correction). This resultant voltage is therefore equal to die change in the error voltage across these two cathodes 2 since the last correction, and a corresponding signal appears across the

output conductors

382 and 384. In this way, the circuit which has been described generates a voltage representing the change of error length between successive corrections.

Referring now to Figure 9, when the relay A operates, the voltage between

conductors

384 and 382 is applied across a capacitor 386 of 1 microfarad, and the voltage between conductors 127 and 127a is applied across a capacitor 388 of 0.25 microfarad. When relay A releases, these two capacitors are connected in parallel across the grids of a further double triode 132, which therefore receives-across its grids a voltage V which is proportional to (Er-KAE) The manner in which the voltage across the grids of the triode 132 is used to control the motor 258 is fully described in our said co-pending application. Briefly the appearance of a signal between the grids of the two symmetrically-connected halves of the double triode 132, which have a common cathode circuit, will result in a As a result, either a relay B in the anode circuit of one half of a double triode or a relay C in the. anode circuit of onehalf of a double triode 149 will operate. If relay B operates, contact B2 willohange over, contactor D will be energised and the motor .258 will, rotate in a first direction. If relay C operates, contact C2 will change over, contactor E will be energised, and the motor 258 will ro ate n t ppo i e,- ire tion. The p riod for which relay B or C is energised and for which the motor rotates, is proportional to the voltage difierence which is applied across the control grids of the double tri'ode 132. As a. result, the ratio of the gearing 254 and the speed of the control draw rollers will be modified in accordance with this voltage difference, and the correction of the register error will commence.

The trigger circuit which includes the triodes 350 and 352 (Figure 8), is also used in the timing circuit for operating relay A. A potentiometer 390 is connected between the anode of the triode 352 and the positive H. T. line, and the wiper of this potentiometer is connected by way of a conductor 392 (Figures 8 and to a differentiating circuit comprising a capacitor 394 and a resistor 396. Negative signals at the junction of the capacitor 394 and resistor 396 are applied through a resistor 398 and a rectifier 400 to the grid of a triode 402. These negative signals also charge a capacitor 404, which is connected between the grid of triode 402 and a point in a potential divider consisting of the

resistors

406 and 408 between the positive H. T. line and earth. The triode 402 is connected as a cathode follower and the diflferentiating resistor 396 is connected to the cathode of this triode. At each operation of the above-mentioned trigger circuit, the grid and cathode of triode 402 become progressively more negative. The output from the cathode is applied through a rectifier 410 to the grid of a first triode 412 which, with a triode 414, forms a monostable trigger circuit of the kind having a common cathode resistor and a capacitor 416 coupling the anode of one valve to the grid of the other. The triode 412 is normally conducting, but when the cathode of triode 402 becomes sutliciently negative the triode 412 is cut off and the trigger circuit reverses. As a result a relay F in the anode circuit of the triode 414 is energised for a period determined by the time constant of the capacitor 416 and a resistor 418.

Contact F2 closes and completes the energising circuit of the relay A, whereupon contacts A1 to A4 change over to charge the capacitors 386 and 388 to the new values of the error signal and the change of error signal, and contact A5 changes over to recharge the capacitor 366. Contact F1 closes to discharge capacitor 404 in readiness for the next charging sequence.

The duration of the period between successive operations of the relays F and A may be adjusted by shifting the wiper of the potentiometer 390 (Figure 8).

By varying the values of the capacitors 386 and 388, the relative effects of the error voltage and the voltage representing change of error can be altered.

We claim:

1. Apparatus for the automatic control of register error of continuous moving webs, including web driving means, correcting means controlling said web driving means and operating periodically to reduce the register error, and means responsive to the magnitude of the error and the change in error during a short period for adjusting the correcting means to vary the magnitude of the applied correction in accordance with both the magnitude of the error and the change in the error during a short period preceding an adjustment of the register correction means.

2. Apparatus for the automatic control of register error of continuous moving webs, including web driving means, correction means controlling said web driving means periodically to reduce the register error, signal-generating means controlled by said web and providing electric correction signals which vary in accordance with the difference between a term representing the magnitude of the register error and a term proportional to the change in the error since the preceding correction, and means controlling said correction means by said correction signals to vary the amount of correction in accordance with said correction.

3. Apparatus according to claim 2, in which 1 Got- 12 rection means includes a compensating roller the movement of which varies the length of a section of the path of the web. H

4. Apparatus according to claim 2, including a pair of draw rollers and in which said correction means includes a variable speed gearbox through which said draw rollers are driven, and means for adjusting the transmission ratio of said gearbox to vary the speed of said draw rollers and thereby reduce a register error.

5. Apparatus according to claim 2 including a printing cylinder and in which said correction means includes means for adjusting the position of said printing cylinder to reduce a register error.

6. Apparatus according to claim 2, including an adjustable brake and in which said correction means includes means for adjusting said brake to vary the tension of the web and thereby reduce a register error.

7. Apparatus for the automatic control of register error of continuous moving webs, including web driving means, means for periodically measuring the register error and generating an electric signal representing said error, means responsive to two such signals to provide an electric signal representing the difference between the two measured error values, means for generating electric correction signals which vary in accordance with both said error signal and said difference signal, electrically operated register adjusting means controlling said web driving means, and switching means for applying said electric correction signals periodically to said correction means.

8. Apparatus according to claim 7, including pulsegenerating means arranged to generate pulses in response to register marks on the web to be controlled, further pulse-generating means arranged to generate master register pulses, electrical means for generating periodically an electric error signal representing the interval between a master register pulse and a pulse representing a corresponding register mark on the said web.

9. Apparatus according to claim 8, in which a magnetic switch is used to generate the master register pulses.

10. Apparatus according to claim 9, in which the magnetic switch also provides gating pulses which are used to remove from the output of the pulse-generating means associated with the moving web, signals corresponding to printed matter between the register marks.

11. Apparatus according to claim 7, including a storage capacitor in which each error signal is stored until the next error signal is generated.

12. Apparatus according to claim 11, in which an electromagnetic relay operates periodically to transfer each error signal to the said capacitor.

13. Apparatus according to claim 12, in which the said relay, when operated, transfers the error voltage and the voltage representing the change of error to two further capacitors, and when restored to its original condition connects these two further capacitors in parallel.

14. Apparatus for the automatic control of register error of continuous moving webs, including web driving means, pulse-generating means for generating master register pulses and pulses representing the passage of register marks on a web, means for measuring the error interval between a master register pulse and a corresponding web register pulse, multiplying means for generating an error signal the amplitude of which varies with the product of the error interval and the speed of the web, means responsive to two such error signals to provide an electric signal representing the difierence be tween the two corresponding error values, means for generating electric correction signals which vary in accordance with both said error signal and said difference signal, electrically operated register adjusting means controlling said web driving means, and switching means for applying said electric correction signals periodically to said correction means.

15. Apparatus accord to cl 14, in which the error pulses are used to block a diode in the grid circuit of an electronic valve and in this way to permit the valve to be controlled for the duration of the error pulse by a voltage representing the Web speed.

16. Apparatus according to claim 15, in which the valve, when rendered conducting by an error pulse, draws current from a capacitor, which is charged to a voltage representing the product of the duration of the error pulse and the speed of the web.

17. Apparatus according to claim 15, including a discharge circuit for the capacitor, the discharge circuit comprising a monostable trigger circuit, operated by the register pulses, the anode of one valve of the trigger cir- 14 cuit being so connected through a rectifier to the said capacitor that when the potential at the said anode is at its lower value during and shortly after error pulses the rectifier blocks and the capacitor is allowed to charge, and when the potential at the said anode is at its higher value the capacitor discharges through the rectifier.

References Cited in the file of this patent UNITED STATES PATENTS 2,250,209 Shoults et al July 22, 1941 2,576,529 McKenney et al Nov. 27, 1951 2,583,580 Ludwig Jan. 29, 1952