US3015974A - Automatic control system for rolling mills and adjustable dies - Google Patents

Description

INV ENTORS Jan. 9, 1962 o. E. oRBoM ErAL 3,015,974

AUTOMATIC CONTROL SYSTEM FOR ROLLING MILLS AND ADJUSTABLE DIES Filed Sept. 18, 1958 3 Sheets-Sheet 1 ATTORN Jan. 9, 1962 Filed Sept. 18, 1958 ORBOM ET AL O. E. AUTOMATIC CONTRO MILLS AND ADJUSTABLE DIES L SYSTEM FOR ROLLING 5 Sheets-Sheet 2 ORNEY Jan. 9, 1962v ORBOM ET AL A O. E. AUTOMATIC CONTROL SYSTEM FOR ROLLING MILLS AND ADJUSTABLE DIES Filed Sept. 18, 1958 3 Sheets-Sheet 3 INVENTORS BY g ` ATTO EY United States Patent Oiice 3,015,974 Patented Jan. 9, 1962 3,015,974 AUTOMATIC CONTROL SYSTEM FOR ROLLING MILLS AND ADJUSTABLE DIES Orville E. Orbom, Brackenridge, James B. Murtland, Jr., Tarentum, and Fred J. Schoepf, New Kensington, Pa., assignors, by'mesne assignments, to General Electric Company, a corporation of New York Filed Sept. 18, 1958, Ser. No. 751,818 3 Claims. (Cl. 80-56) Our invention relates broadly to the production of material by rolling through 'single or multiple stand rolling mills or by drawing through single or multiple dies of the kind having opposing die surfaces whose separation is adjustable and more particularly to an automatic system for controlling such mills or dies so as to produce material of a predetermined desired uniform gage thickness.

One of the objects of our invention is to provide an automatic control system for rolling mills or systems employing adjustable dies which measures gage and controls the mill or dies at the bite of the mill.

Another object of our invention is to provide a control system for rolling mills which greatly reduces the amount of waste material on the ends of a work piece.

Still another object of our invention is to provide an automatic control system for rolling mills which operates on the concept that the volume of material 'leaving the mill is equal to the volume of material entering the mill.

.A further object of our invention is to provide an automatic control system for rolling mills which measures gage of an increment of work before the bite of the mill and stores this information in a memory unit until the same increment of work enters the bite of the mill, at which time the previous gage measurement is 'used for instantaneous mill control.

Still a further object of our invention is to provide an automatic control system for rolling mills which produces material to much stricter and more even .tolerances throughout the length of the workpiece.

Other and further objects of our invention reside in the application of our control system lto bidirectional rolling mills and also in the methods by which the mill correction signals are derived as set forth more fully in the specification hereinafter following by reference to the accompanying drawings in which:

FIG. l is a schematic diagram showing in block form a simplified control system for solving and executing the teachings of our invention; and

FIGS. 2 and 3 form a composite schematic diagram when the right edge of FIG. 2 is disposed adjacent the left edge of FIG. 3, the composite diagram showing in block form a modied and more complex control system for solving and executing the teachings of our invention.

Our invention is directed to the controlledproduction of material by rolling through single or multiple stand rolling mills or by drawing through single or multiple dies of the kind having opposing die surfaces whose separation is adjustable. More accurate dimensional control is an objective of automatic controls for the aforementioned equipment. For simplicity of description, this disclosure will concern itself with the description as applied to single stand rolling mills. Those skilled in the art can obviously see the correlation of the hereafter described systems to other rolling or drawing operations. In the majority of existing control systems for mills or dies, the system is controlled from a gage measurement taken several feet beyond the exit side of the rolling mill, by an elongation or percent reduction measurement, or by a roll pressure measurement` In a system employing gage measurement, the material after reduction progresses to the gage, which may be several feet beyond the bite of the mill, before any error present in material thickness can be detected. This distance from the bite of the rolls to the gage is commonly referred to as transport distance. The time required for the material to reach the exit gage is denoted as Transport Time. Time required to measure the strip gage is referred to as Sensing Time. Transport time andsensing time are major elements in vdeveloping error commands. Transport distances of iive (5) vfeet or more are common in most of the commercial rolling equipment available at present. A system having a transport distance of five (5 feet would not be capable of detecting an error signal until tive (5) or more feet of material had passed from the bite-of the mill rolls. The corrective signal would then be vtransmitted tothe mill screwdown; but, the measuring gage would not detect the result of this action until tive (5) or more feet of material had Vpassed through the mill. From this discussion, it is seen that ,a system of this sort has a natural frequency of oscillation. lf vthis oscillation were left to exist without any attempt to control it, the results would be undesirable. For a material entering the mill with fairly noticeable changes in gage, the system described would cause such wide Variations in output gage that in most probability would eventually result in tearing of the strip. In an attempt to control the natural frequency of oscillation, some designers have damped their system so as to allow only a certain portion of the requested corrective signal after each measurement to be transmitted to the screws. Other designers have provided damping in the screws in order to slow down the response to any corrective signals. In any case, damping considerably reduces the e'ciency and effectiveness of these control systems. In a highly damped system, the result is a number of ensuing measurements before the material is brought within gage limits resulting in the waste of material which has progressed from the mill before the desired gage could be attained. In present day commercial mills. this loss is considerable.

A system employing elongation measurements as a means of control has one basic advantage; the transport distance of such a system is zero if the strip is under tension. Two basic Vtypes of measuring systems may be employed, an instantaneous rate measuring device such as a D.C. tachometer as described in other patents or a sampling type of device as described in Orville E. Orboms copending patent application Elongation Control System, Serial No. 680,349, tiled August 26, 1957. The strip length-measuring devices may be placed at any distance from the roll bite of the mill. In the sampling type system, for a certain increment of length passing through the entering length-measuring device, there will be an equal or greater increment of length measured at the exit device. The exit length is dependent on reduction. Since the transport time (with the strip in tension) is zero, the only time delay in the system will be the time involved in sensing sample lengths at the entrance and exit of the mill. These increments can be made so short as to be ignored. Although it would appear thatl an elongation type of system control would be desirable, further analysis shows a serious shortcoming of this system. An elongation system will only reduce the material a given percentage. This system does not control to a given gage unless the entering strip gage is constant. Since the gage of the entering strip varies, this type of control is an inaccurate means of gage control. To illustrate, let us assume that an increment of material .040 inch thick enters a mill followed by an increment .080 inch thick. It is further supposed that the opening between the mill rolls is such as to reduce the rst increment to a thickness or gage of .020 inch. In doing so, the material must then have been elongated to twice its original length. When the second increment of material (.080 inch thick) enters the mill, it will yalso be reduced to a gage of .020

inch. However, the elongation in this case is four times the original length; therefore corrective mill action will be requested by the system in the second case. But, cor rective mill action is not desired since the exit gage or thickness is in both cases alike. Thus, it can be seen that a system employing elongation measuring methods alone is not feasible for gage control.

Systems utilizing roll pressure measurements as a means of control leave much to be desired when precision control of strip -gage is required. First, an inherent factor in a pressure measuring system is the natural spring of the mill housing. There is no reasonable means of keeping the mill housing rigid under the extreme forces encountered in rolling; Compensating for this housing spring is possible; however, a second disadvantage would still exist in a roll pressure measuring system. This appears as a result of differences in material hardness at different points in a strip. Hard materials will not be reduced as much as soft materials under certain roll pressure conditions.

In the following disclosure we will described a control system based on an entirely new basic concept of mill control, applied in such a manner that the transport time is zero and the measuring time is essentially zero. One of the greatest advantages of the control system of our invention is that it measures and controls a rolling mill at the bite of the mill. Basically, the system is designed on the concept that the volume of material coming out of the mill must be equal to the volume entering the mill.

Thus:

V1= Vg and (1) Y Y L1W1G1=L2W2Gz where L1=length of material entering mill L2=length of material leaving mill G1=gage of material entering mill G-2=gage of material leaving mill W1=width of material entering mill W2=width of material leaving mill If X=the error control signal:

' The broad concept of our invention is any system or combination of systems viz., hydraulic, pneumatic, electrical, mechanical, etc. that will perform the necessary measuring and computations of the above formula so as to measure, compute, and/ or control any material output from any single or multiple stand rolling mill, or measure, compute and/or control any material output from any drawing through single or multiple dies of the kind having adjustable opposing die surfaces.

A In those eases where the lengths, widths, or thicknesses may remain constant, this formula may be simplified into several simplified versions.

In cold rolling, width of strip usally remains constant even though the material is being passed through the mill. Some materials to be rolled will be encountered wherein the exit width will not equal the entering width because of rolling forces. If it is found that there is no difference in exit and entering widths, then W1= W2 and Formula 1 can be simplified to:

For the case where the exit width differs from the entering width by a known constant amount, a constant correction can be applied to any system designed on the basis of Equation 2 to compensate for this difference in widths. In systems where the width is continually varying, width measuring devices can also be added to reflect the variation to equation solving equipment. In claiming a system capable of measuring and controlling at the bite of the mill and eliminating transport distance, it is our intention to apply the above equation in a manner wherein the input gage (G1) will be calculated on the basis of the other two or three variables (L1, L2, and G2), then compared against the actual measured value of G1. Equation-wise, this may be written:

L G1 :GZ

where @1 represents calculated input gage.

X: G1 l-' G1 %G2d where G1 represents measured input gage Gzd represents desired exit gage As discussed previously in the description of the lelongation type of control system, transport time is zero, and the incremental sampling length can be made so short as to allow the elongation or length sensing time to be ignored. This feature is an integral part of the system of our invention due to the elongation or length measuring means (L1, L2) which we employ. In addition, the desired exit gage (Ggd) is used in the calculation of desirable input gage, rather than actual output gage (G2); thus, the control system of our invention is capable of control before the fact rather than after -the fact. In other words, the system will not need to wait until the material has progressed beyond the bite of the mill before recognizing and deciding on the necessary control operation. Our system -anticipates and recognizes the necessary mill screw action as the material approaches the mill.

It should be remembered that any number of solutions to the above equations are possible; and this disclosure will therefore only attempt to illustrate a few in the following examples. The inventors desire that it be understood that thefollowing examples be considered in the illustrative sense rather than in the limiting sense and that our invention lies in the following basic concepts:

(1) Control of a system based on constant volume.

(2) Control of a system based on constant volume, and elimination of transport distance and transport time by calculation of desirable entrance gage (51) from the variables of entering strip length (L1), exit strip length (L2), and desired exit gage (Gm).

(3) Control of a mill based on L1W1=L2W2, where =G2.

(4) Control of a mill based on L1G1=L2G2, where WlZWg. i

(5) Control of a mill based on W1G1=W2G2, where L1=L2.

(6) It is possible to control thickness of product by this system, as set forth in the following examples.

(7) It is possible to control width by this system.

(8) It is possible to control length by this system.

The following will beV two specific solutions to control a rolling mill and will serve as rm examples of types of equipment that might be used to solve the above equations and control the mill.

The discussion following is based on the example system shown in FIG. 1. In applying a digital solution to Equation 2, the equation will be used as written. However, the factor G2 (mill exit gage) will now represent the desired exit gage rather than the calculated gage. This is permissible in our constant volume control system since the exit length (L2) when used with the entrance length (L1) and entrance gage (G1) actually reects exit gage. The digital system will not only compare the terms of Equation 2, but will actually solve for the amount of error in the comparison. In other words, the digital system will solve the constant volume Equation 2 in the form:

lOf. course this equation would always have zero as a nal result if the exit gage (G2) was the actual measured or calculated quantity (as long as there was no deviation in width or as long as a correction was applied for width change). Since in our case, we inject the desired exit gage (constant quantity) into the system to represent G2, an error factor will vappear in the final solution of Equation 2 at any time when the factors L1, G1 and L2 do not appear in proper proportion to combine with G2.

Thus the system solution of Equation 2 can be written in the form Y A simplified schematic diagram of a typical system illustrating the equipment necessary for digital solution of our error equation is shown in FIG. l.

Reference character 1 represents a conventional rolling mill having pressure rolls 2 and 3 between which pass the material 4 being acted upon or processed in the mill which feeds olf of payoi reel 5 and is coiled onto take-up reel 6. Entrance gage is measured by a digital type displacement gage 7 with an attached binary digitizer 8 to convert gage information into switch closures; thus, a coded readout of entrance strip thickness will be procured. The electronic ambiguity contro '9 is a device for removing ambiguity common to the two brush binary digitizer. This ambiguity is such that in a case of an output progressing from a count of 63 to 64, any combination of the 25-25-24-23-22-21 relays may be received, whereas only the 26 relay should be energized. Therefore an ambiguity control unit is necessary. The ambiguity control unit 9 need not be a relay unit as ordinarily accompanies most commercial binary digitizers but can be an electronic type of ambiguity control which gives faster response than relay type controls. It should be noted that the output of the binary digitizer `and ambiguity control `approaches 211 binary units or actually the maximum is 2047 decimal units. This output is necessary since it is desired to measure entrance gage to a maximum of 0.2000 inch. The binary digitizer 8 and electronic ambiguity control 9 together perform the function'of converting the analog output from displacement gage 7 to a binary representation of the analog output. The book Digital Computer Components and Circuits, by R. K. Richards, published in 1957 by D. Van Nostrand Company, Inc., contains an extensive chapter 1l, from page 459 to 501, on Analog-to-Digital and Digital-to-Analog Computers plus an extensive bibliography on pages SG1-503. A suitable 'two brush binary digitizer plus an ambiguity control is described in detail starting on page 471 of Richards.

The gate 10 acts as ya switch controlling the storage of the measured entrance gage (G1) in the magnetic memory 11 designated as Magnetic Core Shift Register Matrix. Normally the gate 10 is closed and only at predetermined intervals preset in Ithe G1 Storage Counter 12 will the gate open. A suitable gate is described on pages 107- lll of Richards.

As mentioned before, the magnetic core shift register matrix 11 is the memory or storage point for the measured entrance gage (G1). Primarily the memory is composed of ten storage chambers; each storage chamber having eleven magnetic cores to represent the progressive binary digits which must be combined to represent a certain measured gage. Automatic shifting is provided to remove a stored gage measurement once the information has been used and is no longer necessary or desired. The automatic shift circuitry also shifts the stored information in one chamber to the next succeeding chamber. Thus the entrance gage information in the storage chamber most distant from the gate 10 is the gage measurement for the incremental strip length about to enter the mill rolls. Perhaps it should be explained why 6 ten storage chambers are provided. Ordinarily, the entrance strip gage must be mounted about tive feet from the bite of the mill. Since extreme variations in gage are not normally encountered, a measurement of strip gage six inches of strip length will be suilicient to reveal deviation in gage. On the basis of a measurement over each six inches of length and adistance of live feet between the mill and the gage, a magnetic core storage system with ten chambers will be necessary. More memories can be added if closer control is desired; for example, 20 memories and 5 feet from gage to roll bite would give new values every 3 inches. A suitable magnetic core shift register matrix is shown in U.S. Patent 2,946,047 to Morgan. The memory shown in this patent has only two n storage chambers but it is well withiifthe skill of one skilled in 4the art to enlarge the memory toten storage chambers as called for herein. In the patent each storage chamber has only eight magnetic cores but again it is Well within the skill of one skilled in the art to enlarge each storage chamber to ten magnetic cores. In the patent the inform-ationis shifted into the rst storage chamber serially and transferred to subsequent storage chambers in parallel.

Input strip length (L1) is sensed by equipment similar to that described in the copending application Elongation Control System, Serial No. 680,349, O. E. Orbom, led August 26, 1957, namely a pulse generator 13 and demodulator unit 14, and a pulse counter 15. The pulse generator 13 is coupled to idler roll 16 on the entrance side of the mill, and produces tive hundred electrical pulses per generator revolution. Each generator revolution will represent a certain length of strip. The entrance strip length will be preset into the base counter 15. As shown in FIG. 1, the base counter can be preset to count in strip intervals -between 128 andl6,384 pulse counts. When the pulse count at the base counter 15 reaches the preset value, a pulse of electrical energy will be transmitted to the L2 Storage Chamber Gates 17, to the Output Counter Reset Delay Chamber 18, and to the G1 Storage Counter 12. A book, Pulse and Digital Circuits, by Jacob Millman, published by McGraw-Hill Book Company, Inc., in 1956, describes on page 345 a commercially available preset counter which may be used as the preset counter base counter 15. Also see reference note 9 on page 353. In supplying a pulse of energy to the L2 Storage Chamber Gates 17, a command is initiated for the count of exit strip length (L2) to be stored. This is accomplished by opening the L2 Storage Chamber Gates 17, and allowing the count from output counter 19 to pass into the computer L2 Storage 20, which is primarily a magnetic core memory. Patent 2,946,047 to Morgan again describes a magnetic core storage register. It is well within the skill of one skilled in the art to adapt the magnetic core storage register disclosed in Patent 2,946,047 to the required number of magnetic cores. Exit strip length (L2) is sensed by pulse generator 21 coupled to idler roll 22 on the exit side of the mill and produces a series of pulses with the aid of demodulator unit 14 in the same manner as the L1 pulses were produced, and transmits them to output pulse counter unit 19. Patent 2,848,166 to Wagner describes a suitable counter and associated gates for output counter 19 and L2 storage chamber gates 17. At the same time, the same energy pulse transmitted from the base counter 15 to the Output Counter Reset Delay Chamber 18 triggers -a delay circuit which after a speciic elapsed time will reset the Output Counter 19 to zero. Suitable delay circuits for delay circuit 18 are described on pages lOl-103 of Richards, Digital Computer Components and Circuits. The G1 Storage Counter 12 will also receive one pulse of energy from the base counter for each unit length of entrance strip. G1 Storage Counter 12 is preset to allow only one storage of entrance gage G1 for several measurements of entrance and exit length. Again, a suitable preset counter is referred to on page 345 of Millman, Pulse and Digital Circuits. Entrance gage (G1) would not show any appreciable deviation if it were measured and stored during the same short intervals that entrance length ismeasured over. Therefore, it is only necessary to store one entrance length measures. It should be remembered, however, that the gage is being measured continually even though it is only stored at preset intervals. In a typical case, entrance length may be measured in terms of one inch increments whereas entrance gage measurement and storage may only be desirable for each six inches of strip. Therefore, the G1 Storage Counter 12 will be preset to count six pulses before initiating a command for storage of entrance gage. As can be seen in FIG. 1, when the pulse count at the G1 Storage Counter reaches the preset value a pulse of energy will be directed to the magnetic core shift register matrix 11 to cause the entrance gage value in the last storage chamber (closest to computer) to be destroyed, and then to cause the information in the ensuing chambers to be shifted by one chamber. While this is occurring, energy from the G1 Storage Counter is also delivered to the delay'unit 23 and after a specic time interval, the delay circuitry will allow energy to ow to the gate -to open the gate and allow the measurement of entrance gage G1 to be stored in the magnetic core shift register matrix 11. A suitable delay circuit for delay circuit 23 is described on pages lOl-103 of Richards, Digital Computer Components and Circuits. Pulses owing to the gate 10 also ow to the Control Initiating Counter 24. The purpose of this one preset counter is to delay control of the mill at start up until the magnetic core shift register matrix has been completely lled with entrance gage measurements. This is necessary since the entrance gage information entering the magnetic core shift register matrix is stored in the first chamber and shifts one chamber at each ensuing length measurement; thus, a particular entrance gage measurement is not used until it has reached the tenth and final storage chamber in the magnetic core shift register matrix. It can be seen that this arrangement also insures that the proper entrance gage information is associated with a particular increment of strip length when the calculation for the increment is made immediately prior to its entrance into the bite of the mill. In our case, it was assumed that ten increments of strip length would be measured before the rst increment entered the bite of the mill. This would mean that the control initiating counter 24 would be preset for ten counts at which time it would close relay contacts and connect the common side of the power source to the output relays in the mill control output unit 25. Since the output relays in unit 25 must be energized when a mill correction is desired, no mill control will .be initiated until the control initiating counter 24 has counted to the preset number. Again a suitable preset counter is referred to on page 345 of Millman, Pulse and Digital Circuits.

The width offset 26 is provided for correction of the system to compensate for any deviation of the strip width from the initial assumed constant value. The correction is injected into the base counter by triggering the base counter to a certain count value at the beginning of each entrance strip measurement. In other words, a typical correction for an entering strip increment of 100 counts with a 10% deviation in width would be l0 counts. Thus the base counter would be reset to -10 at the time that each new entrance increment was begun to be measured. This is accomplished with a series of switches in the width offset unit 26. The switches when closed will supply bias voltage to certain tubes in the base counter, and force the tubes to a count 10 condition at reset. The switches can be set manually or if width measuring devices yare used, the switches will be electrical in nature and can be controlled from the width measuring devices.

Desired exi-t gage (G1) is injected into the system with the G2 desired gage decimal switches 27. The apparatus 27 consists of a single preset decimal counter, an oscillator and a series of pushbut'tons. The operator selects the desired gage on the push-buttons. Circuitry on the pushbut-ton contacts is so arranged that the complement of the desired gage is actually reflector to the single preset decimal counter; in other words, if the operator inserts a pu-shbutton gage selection of .0200 and the maximum gage that can ever be attained is .2000, the pushbutton circuitry actually reflects the complement of the desired gage (.1800). Thus the single preset decimal counter is triggered by the pushbutton circuitry toV a count of 1800. After inserting the desired gage, the operator then pushes the start button 28, at which time the oscillator in the G2 desired gage decimal switches unit 27 will begin to oscillate and send pulses of energy to the single preset decimal counter, as well as to the pure binary counter 29. The single preset decima-l counter is -arranged to stop the oscillator when the single preset decimal count reaches -the maximum of 2000 units.. Since, in our assumed case, the single preset decimalV counter was initially triggered to an 1800 count, the os cillator will send out pulses until the single preset decimal counter reaches a count of 2000, yat which time the os-v cillator cutoff will occur. Since the initial single preset decimal counter count was 1800, and the final count was 2000, only 200 pulses of energy were received. These 200 pulses of energy were Aalso transmitted to the pure binary counter 29, but this counter had started at zero, therefore, it will display a count of 200 at shutoff. This was the initial gage set into the pushbutton. At shuto, the count attained in the pure binary counter 29 will be transmitted to the ,G-g multiplier 30 in the computer. This count will be in binary form. Patent 2,848,166 to Wagner again shows and describes a suitable binary counter for pure binary counter 29.

Each time the operator inserts the desired exit gage and pushes the start button 28, a pulse of energy will be sent totheV Control Initiating Counter 24 to reset it.

VOnce the individual functions of Equation 2 have accumulated in the system, the following actions occur:

(l) The entrance strip gage (G1) associated with the first increment of strip length, and stored in the magnetic core shift register matrix 11 tenth chamber, is multiplied by the leng-th of incremental strip (L1).V This is accomplished in the L1 Multiplier unit 31 and the nal result is a product (L1G1). It should be mentioned that the ,L1 Multiplier unit 31 is equipped for manual insertion of the L1 value. A series of rotary switches is provided with which this adjustment can be made. As described on page 11, base counter 15 is preset to the number representing the strip interval L1 which is to be multiplied in L1 multiplier 31. When the counter 15 has counted a one strip interval L1 the base counter 15 transmits a pulse to G1 storage counter 12 indicating that one increment L1 has been measured. Storage counter G1 has been preset to six Iand after six increment lengths L1 have been measured and base counter 15 has transmitted six pulses -to G1 storage counter 12, G1 storage counter 12 counts to the preset number of six and transmits a. pulse to the magnetic core shift register matrix 11 to cause the entrance gage in the last or tenth storage chamber to be transferred to L1 multiplier 31 and the entrance gage in the other storage chambers to be shifted to the next succeedingstorage chamber. As the input gage G1 was measured at six inch intervals with the measurement made iive feet from the rolls as described on page ll, the entrance gage stored in the last or tenth storage chamber is the gage of the increment length L1 which was counted in base counter 15 six times to cause the shift in the magnetic core shift register matrix 11 and the transfer of the gage to L1 multiplier S1. The incremental length L1 has been manually inserted into L1 multiplier 31. Thus the transfer of the entrance gage G1 into the L1 multiplier 31 results from the measurement of Van increment length six -times and the entrance gage G1 is automatically multiplied by the increment length L1 in the L1 multiplier 31.

(2) Exit strip length (L2), 'as stored in the L2 Storage magnetic core memory 20, is multiplied by the desired exit gage G2 and a product (LzGz) is obtained. This multiplication is accomplished in the G2 Multiplier 30. This unit is similar to the L1 Multiplier 31, except that multiplier insertion is not manual but automatic. The multiplier value is ladjusted as a result of switch closures in the Decimal and Pure Binary Counter 29. The L1 multiplier 31 and G2 multiplier 30 may `consist of digitalto-analog converters for converting the digital data in-to analog form and analog multipliers. In L1 multiplier 31 the insertion of the L1 value may be directly in analog Y form. Richards, Digital Computer Components and Circuits, in chapter 1l, pages 459-501, describes digitalto-analog converters that may be used to convert the digital information into analog form. Analog multipliers suitable Ifor multiplying the analog values are well known to those skilled in the art with such an analog multiplier described in Patent 2,735,616 to Hoadley.

(3) Once the products LlGl and LZGZ are obtained, the smallest value must be subtracted from the largest value to obtain the difference. This is done in the Subtractor 32. Millman, Pulse and Digital Circuits, in chapter l5, pages 459-485, describes voltage comparators in detail which may be used as the Subtractor 32. Perhaps it should be mentioned that the product LlGl will be represented by a voltage of certain polarity and magnitude While the product L2G2 will be represented by a voltage of opposite polarity but substantially the same magnitude. Thus the products can be compared to determine which is the larger of the two. When a difference in value exists between the two products, we have then detected that an error in exit gage will occur if the mill is not adjusted for the strip increment about to enter the bite of the mill. Since it is advantageous in most cases to tolerate a small amount of error, allowance is made in the subtractor to provide a manual control for adjusting the amount of error that can be detected without any ensuing adjustment to the mill.

(4) Once a signiicant error has been detected in the Subtractor, an adjustment request will be transmitted to the mill via the Mill Control Output unit 2S. The Mill Control Output unit 25 consists of two relays. One relay will upon energization start the mill screws represented at 33 upward, `while the other relay will cause the mill screws 33 to move downward. The particular relay energized will be determined `by the polarity of the error detected in the Subtractor unit 32. Screw-up or screw-down operation which is controlled through Screw Control unit 34 will continue as long as a significant error is detected at each calculation and transmitted to Screw Control unit 34 from Mill Control Output 25.

FIGS. 2 and 3 illustrate a second method of rolling mill control based on the previously mentioned Equations l and 2. As can be seen from the diagram, many of the component parts are the same' parts as appeared in FIG. l. For purposes of further iden-tiiication, each major component will be briefly discussed below before attempting an overall system performance description.

Reference characters 3S and 36 represent thickness or gage sensing devices of the displacement, X-ray, gamma ray, beta ray, etc., type. FIG. 2 shows the gage sensing devices, also labelled GL and GR for the reason that the direction of rolling will determine whether the left gage (GL) or the right gage (GR)` output should be used to represent the input gage G1 of the mill.

Reference characters 42 and 86 represent servo type followers which accept the gage readings (usually in electrical form) of components 35 and 36 and converts them into mechanical shaft rotations represented at 43 and 87 respectively.

Shaft position digitizers 44 and 88 convert the shaft rotations 43 and 87 representing material thickness (as reflected in components 42 and 86) into a mathematical form suitable yfor computer input. The mathematical readou-t may be either of the decimal or binary class. However, in this case, the binary type of digitizer will be assumed as being used herein.

Decode Logic unit 45 is an electronic repeater of the digitizers as well as a device which removes the ambiguity common to most two brush commercial binary digitizers. The unit also includes means for switching outputs of digitizers 44 and 8S in and out of the output channel connecting to the magnetic core buier 46. Depending on the direction of rolling of the mill one of the digitizer readings (and only one) will be submitted to the magnetic storage chamber 47 via the magnetic core buffer 46. The direction of rolling will be selected by the mill operator at the mill drive direction selector unit 55, and a reiiector of this selection will be relayed to the decode logic unit 45 over circuit 85. The decode logic unit will then make the proper switch contact closures to insure that the proper digitizer output is supplied as an indication of mill input gage to the magnetic storage memory.

The magnetic core buifer unit 46 is simply a buier ampliiier inserted for attaining better impedance match between the magnetic core load (of the magnetic core memory) and the transistor repeaters of the decode logic unit 45. Gain in stability is procured by this design since vthe buffer .amplifier is usually designed to draw no grid current. Also, included in this unit is an electronic gate or latch which only allows the input strip Ygage measurement to be stored at certain intervals.

The magnetic core memory 47 is the storage point wherein the measured material entrance gage G1 is contained until compared with calculated value. Primarily the memory is `composed of ten storage chambers; nine of these chambers each have eleven magnetic cores to represent the progressive binary digits lwhich must be combined to represent a certain measured gage. The tenth storage chamber is consuucted of transistors rather than magnetic cores, since it is necessary to read-out the information from this chamber in serial fashion. Magnetic `core ymemories do not lend themselves to serial fashion readout conveniently.

The strip length sensing means are the pulse generators 13 and 21 coupled to the idler rolls 16 and 22 respectively of the mill. These pulse generators receive an RF. signal from the demodulator units 14 and modulate it according to strip length passing each length sensing element.

Demodulator units 14 are a source of RF. energy for the pulse generators 13 and 2l, as well as a means of reconverting (or demodulating) lthe modulated signal into pulses of D C. energy representing strip length. Thus the combination of a pulse generator and its respective demodulator unit will produce either 500, 1000 or 2000 electrical pulses per revolution of .the pulse generator, each revolution of the pulse generator in turn will represent a certain strip length of material entering the mill.

The system of this disclosure concerns a control means not only for unidirectional rolling mills, but also for reversible mills, therefore, the electrical pulses representing strip length from the entrance and exit sides of the mill must be directed to the proper pulse counting equipment at each reversal of the mill. In other words, the mill entering strip length pulses must always be recorded in the L1 counter 59 and the mill exit length pulses in the L2 counter 60. To do this the switch 54 is supplied. Each time that the mill operator changes the mill drive direction selector 55 it will relay this information to the vswitch 54 (over circuit 56) and cause it to reverse lines A and B from the demodulators 14.

Multiplier 57 serves asa sensitivity selecting means. By use of it, selectionjcan be made as to whether each revolution of the pulse generators 13 and 21 will result in 500, 1G00 or 2000 electrical pulses being produced. This is accomplished lby the use of only one of the etched patterns in the pulse generator (of the type, for example, as produced by Telecomputing Co.), by the use of both of the etched patterns or by the utilization of both the negative and positive pulses produced in both etched patterns As mentioned before, the equipment of this invention is not limited to unidirectional mill control. But each time that the controlled mill is reversed, certain control system components must be preselected to take care of this reversal. It is the prime duty of the mill drive direction selector, unit 55, to perform this function. This unit supplies a `direction signal to cause contact closures in units 45, 54, and 61 by means of circuits 85, 56 and 64, respectively. Y

In discussing Equation 2, described previously in column 3, mention was made that correction can be applied to this type of solution if it was noticed that some of the supposed constant factors were found to be in error. One of these errors may be encountered due to slight physical and operational differences in the pulse generators 13 and 21. An error constant in magnitude, and quite significant in value is encountered when the pulse generators 13 and 21 are driven from discs designed to ride on the mill idler rolls v16 and 22. After some time, the discs will show wear or differences in diameter, such that if the mill were to be opened to produce no reduction and a certain pulse count increment submitted from one side of the mill, the pulse generator on the opposite side of the mill would not produce an exact same pulse count increment. Unit 61 is the means for manually settingv in this correction. Since the correction is dependent on mill direction, two adjustments. left and right," are provided.

Error caused by a change in width of strip can be compensated for by means of component 62. The W factor is introduced by rotary selector switches into the Width Factor Unit 62.

The error constants K and W, as discussed under components 61 and 62 must be multiplied and then applied to the desired output gage G2 and elongation ratio in order to calculate the desired value of G1. Mathematically, we

say:

Epi: GZWK The multiplications shown above can be performed in a number of Ways, and it is not intended to limit this invention by specifying only one means of doing this; any num- WK: WX K We will now alter the form and say K: 1 |AP W: 1 +AW therefore AK=K l AW: W- 1 then The last equation is of the form utilized in component 81.

multiplying equipment uses only the digits 18 and 50, rather than 1018 and 1050. VThis also results in a saving of equipment. The equipment contained in component 81 includes an oscillator, several electronic gates and a number of pulse counters. Equipment arrangement to effect this multiplication will not be discussed here since the technique utilized is a standard computer technique.

The desired output gage selector, unit 63, consists of a manual selecting means (such as rotary switches), oscillator and a counter. Mill operators can inject the desired output gage into the unit, wherein it is registered as a pulsevcount on the counter.

Unit 66 is very similar to unit 62 since a multiplication type function is desired. The desired output gage G2 and the product of the correction factors (in the form WK-l) is multiplied in this unit. As can be seen from FIG. 3, the result is G2WK, which is then relayed to the L1 counter unit 59, to serve as a preset for this counter.

Reference character 65 designates a pushbutton with which the mill operator initiates the solution of the product WKGZ, clears old information from the memory and starts introduction of new G1 values into the memory unit.

The pushbutton closure (component 65) is taken as a notice by command unit 75 to clear or reset counter 77 and to trigger oscillator 76 into a pulse conducting state. At the same time gate 78 is commanded (over circuit 89) to open and transmit pulses from oscillator 76 (via units 78, 79, 52 and 46) for the purpose of clearing (by stepping) the magnetic memory of any old information that may be present, as well as supplying the memory with control signals for the storage of ten new readings of material gage. It may be noteworthy to mention that This form saves computation time since the factors on the right hand side of the equation deals with fewer significant digits than the W and K factors originally contained. Thus with a case Where K:l.0l8 and W:1.050, the

the strip increments over which Vthis gage information is accumulated will not necessarily be of the same length as the gage measuring increments later determined by the shift preset counter unit 51. Forrstart up, however, this condition is found to be satisfactory.

A further duty of unit is to hold dual gate unit 58 in the closed state until the magnetic memory is lled with input gage information. This is yaccomplished by not allowing electrical energy to ow to the end gate" Iunit 83 (via circuit 84) until the magnetic memory has reached the full state, and also the GZWK multiplication has been completed.

Unit 76 consists of a standard electronic oscillator, controlled from unit 75, and transmitting pulses to units 77 Iand 78.

The counter 77 is also a standard unit of the preset counter nature. In this application, the counter is preset to ten counts, at which time, it closes the gate 7S to announce the completion of the input gage storage in the magnetic memory.

Gate unit 78 is of the electronic type also, and serves :as a controlling and transmitting means of oscillator pulses used to step or command the magnetic memory storage circuits. As previously mentioned, this gate is opened (allowed to transmit pulses) by a signal from unit 75 via circuit 89 and the gate is closed by a signal from the ten-count counter 77.

Shift preset counter unit 51 is the means of commanding the magnetic memory 47 to make a gage storage, as well as, commanding the memory to shift all gage information forward one chamber. The shift preset counter is only used in this capacity after the initial start-up of the automatic control system. As mentioned before, at start-up the units 75, 76, 77 and 78 control the magnetic memory 47. After the initial start, those units are no longer used, and the shift preset counter unit 57 controls all storage in the magnetic memory.

Previous discussions brought out the fact that extreme variations in gage are not normally encountered and that normally one reading of gage is stored in approximately six inches of strip. This length is adjustable in the preset means of unit 51. Since the storage command is related to certain increments of entering strip length, the shift preset counter unit 51 receives its input information from the mill entrance length sensing line L1. This information is -inthe form of pulses, which are then counted in the shift preset counter. The existing relationship between the number of pulses generated by the pulse generators 13 and 21 as compared to the actual length of strip passing into the mill during the pulse generating time, as well as the length increment over which gage is to be measured, will determine the count at which the unit 51 should be preset. When the count reaches this value, a command (pulse) is given to store another reading of input gage G1. This command is routed through the components 79, 52 and into component 46.

The or gate unit 79 is a transistor circuit arranged in standard electronic fashion so tha-t output energy will be supplied to the magnetic core drive amplier unit 52 at any time that input energy appears on either circuit 9i? or circuit 91.

A magnetic core drive amplier 52 primarily of the buier amplifier type, is necessary in order to supply a low impedance pulse for driving the magnetic storage equipment. Unit 52 makes this low impedance pulse possible. In addition, this unit furnishes a lock circuit which serves to delay any read-in (storage) signals during a subtraction operation. This is necessary, other- Wise a shift of gage infomation from one chamber to the following chamber may occur, and thus the information in the tenth and nal chamber (being used in the subtraction operation) may be destroyed. The signal to lock is given by the subtractor unit 48 over the circuit 53.

And gate unit 83 is of Ithe standard transistor variety described under or gate unit 79. The prime function of this unit is to act as a gate in controlling electrical energy flow to the dual gate unit S to allow it to open (via the or gate unit 70). Since this -unit is of the and gate type, energy will only be delivered to unit 70 when the two input circuits 84 and 82 have signals appear on them simultaneously. The significance is that the automatic control system has beenput into operation, ten storages of input gage have occurred, and that the multiplication of the desired output gage G2, width factor W, and correction factor K has also been completed.

Or gate unit 7i) is also -a standard transistor circuit arrangement as discussed under or gate unit 79 and and gate unit 83, however, it will produce Van output suitable for opening `dual gate (unit 58) when a signal appears on either of its two inputs, circuit 92, or circuit 69. The signal controlling conditions or circuit 92 have been discussed under unit 83. As to circuit 69 a signal will appear on this circuit after each transfer of @1, from unit 60 to unit 68.

Dual channel gate 58 is a vacuum tube unit so constructed that `an electrical signal from circuit 93 will cause the tubes to conduct and pass the signals on the input channels. A signal on circuit 67 will cause the gate to close and stop passing of the input channel information. The signal controlling factors of circuit 67 will be discussed in the description of component 59.

L1 counter unit 59 consists of a binary single preset counter. The value to which this counter is preset is determined by the product GZWK multiplied by a binary shift point constant (some power of two). The L1 counter receives electrical pulses (representing strip length) from the dual gate unit 58. When the pulse count reaches the preset value, coincidence circuits respond to close dual gate unit 58 (to block all length measurement pulses from proceeding to the L1 and L2 counters), to notify L2 counter unit 60 to stop counting and transfer this iinal count to the readout unit 68 and Vto notify the readout unit 68 to store the count reading of the L2 counter unit 6G. As soon as the L2 count has been transferred and stored in the readout unit 68, a signal will be transmitted from the readout unit 68 to the L1 and L2 counter units 59 and 60 to cause the counters 14 to reset to zero, also the same Asignal will be conveyed to the dual gate unit 58 via circuits 69 and 93 to cause the gate to open and again allow pulses to ow to the counters.

In the opening statement of this explanation of L1 counter unit 59, brief mention was made that the value GZWK was multiplied by a binary shift point constant before being used as a preset for L1 counter. This binary shift point mechanism (considered part of the L1 and L2 counter hardware -arrangement of FIG. 3) consists of -a stepping switch home-seeking device, in conjunction with a manual selector, the combination of which operates to multiply GZWK by some power of two (as selected with the manual selector). The manual selector -in this case is of the three position type, and re'ects multiplications by -a factor of 20, 21, or 22. Gain in accuracy is the main advantage of this device. This may be accomplished in two ways. First of all, it is common knowledge that the count of L1 counter unit 59 is more significant and accurate if this count is within the onehalf to full scale range of the counter. Thus, multiplying a small value of GZWK by 21 or 24 to bring the present value of the L1 counter Within the above mentioned range would be advisable. Secondly, the increased preset value of L1 counter will bring about a longer measuring time and possibly :a better measurement of L2 Circuitry from a second contact bank on the stepping switches is used -to automatically shift the binary set point in the reverse direction on the count -attained in the L2 counter unit 60. As an example, if the GZWK -value is initially multiplied (with the binary shift point selector) by -a value of 21, then the count of L2 counter unit 60 will also automatically be divided by 21.

Unit 60 -is an electronic binary counter unit similar to that of unit 59, except that it does not include any preset control. The ou-tput material length pulses are counted and upon command `are submitted to the readout unit 68V for storage; the acceptance of the information at the readout unit then being followed by a reset signal to the L2 counter unit 60, and the unit then again being ready to count pulses.

It is desirable at this time to explain how the division operation in our solution for the desirable input gage 1 has been performed. As previously written:

unit 59, the GZWK value was used to represent the preset value of L1 counter, or in other words, L1 was set equal Thus, the L2 count is an absolute reflection of the calculated desirable input gage.

Readout unit 68 consists of `a series of transistor flipops circuits so arranged as to store the L2 count in binary form. Each ilop-op circuit represents a certain ybinary digit. Gating circuitry is provided to allow transfer of information in and out of the equipment only at proper intervals. This gating circuitry is controlled by circuits 67 and 94.

The subtractor unit 48 is constructed of standard transistorized logic circuits which serially (one bit at a time) compare the measured value of input gage G1 (as stored in the tenth chamber of unit 47) against the calculated desired value of input gage 1 (as stored in unit 68). Here again, transistor flip-flop circuitry is utilized to store the difference between corresponding bits. This information (magnitude of difference *between G1 and l,

15 as well :as the sign of the difference) is then sent to the mill control unit 39 (via units 73 and 74), where it is set -into nip-flop circuits controlling the `alarm and screw operations.

IIn order to synchronize all operations, the subtractor unit 48 has a built in distributor. The distributor provides the timed signal necessary to control the readout of G1, Cil, also to control the subtractor and the dead zone and alarm set unit 73, and to reset unit 68 by means of circuit 94. The composite mechanism of this unit consists of a multivibrator transistor pulse gates and transistor nip-nop sequence circuit. As the gating circuits and Hip-flop sequence circuit respond to the successive incoming pulses, signals are transmitted to the various components of the control system, viz., a signal is given for the first binary bit in the tenth chamber of unit 47 to be ytransferred into the subtractor by circuit 50. On the next ip-op action, a signal is given to unit 68 (via circuit 94 to have the corresponding binary bit transferred to the subtractor, ete).

Dead zone and alarm set unit 73 consists of a manual selec-ting means (2 sets of 8 pushbuttons each) for selecting the amount of deviation between G1 and 'G1 that can occur before any mill control -is initiated. This rst -set of pushbuttons is circuited to the and gates and flip-nop circuits of the mill control unit 39, yand determines when and whether a screw-up or screw-down operation should occur.- The second set of pushbuttons determines (over similar circuitry) the maximum deviation which can bet'oleated. This means is used to energize alarm relaysin unit 39 which in turn discontinue automatie control. on the mill screws.

l Although the relay controls vcan be energized once a significant error has been realized, the time a relay may stay energized is controlledby the delay circuit of the time control unit 74 The delay circuit is simply an R-C circuit (with adjustable resistance) in which the charge rate of the timing capacitor determines the length of time that associated transistors pass current; this, in turn determining when a reset signal should be 'given to the relay control flip-flop circuits of unit 39.

As explained previously, the mill control unit 39 consists of several and gates, ya series of transistor flip-flop circuits and four relays (LO, MLC, MHI, HI). The and gates upon receipt of simultaneous signals (at a time determined by the distributor) will set the flip-nop circuits if a difference equal to or greater than the preset values of unit 73 was obtained in the subtraction process. The state of the dlip-op circuits then determines which of the above mentioned relays is to be energized. Energization of the LO or HI relays signifies a difference greater than the permissible maximum has been obtained; an alarm yis sounded and la mechanically latched relay is energized; The mechanical latch relay breaks all control circuits -between the control system and the mill screws. The mill must then be manually controlled.

, Return to automatic control cannot be made until the mill operator removes the mechanical latch by pushing a reset button.

Energization of the MLO or MHI relays closes circuits to the solenoids on the mill screws in screw control unit 34. Depending on which relay (MLO or MHl) has been energized, the mill will respond with either screwup or screw-down action to produce material on the output side of the mill with constant gage thickness. As lstated previously, the length of the screw-down or screwup operation is controlled from the time control unit '74.

System operatz'on.-Reference character 1 schematically designates a conventional rolling mill having coact- ,ing pressure rolls 2 and 3 through which the material 4 being engaged passes from the coiled roll 5 to the coiled roll 6. For the purposes of this discussion we will .considerthat the material being rolled is passing from coiled roll 5 to coiled roll 6. Thus we iind that 16 the gage GL in position 35 is on the input side of this roll and GR in position 36 refers to the gage on the exit side of the roll.

When the material is traversing the mill in one direction measurements are effected by one gage, whereby when the mill is reversing and operates in the opposite direction the measurements are effected by the other gage. The gage measurement G1 in each case, which is used for purposes of control of the mill, is the gage measurement `on the entry side of the mill, regardless of direction. It will be understood that the invention is applicable to both unidirectional and bidirectional mill operations and that in the instance of unidirectional mill operation gage -measurement is effected only on the entry side of the mill.

The mill rolls 2 and 3 are controlled by various means which, in this instance, has been indicated as a screw 33 operated by screw control 34 through leads 37 and 38 from the mill control designated at 39. The mill control 39 has been shown as including the screw-up circuit 38 and the screw-down circuit 37 and separate alarm control circuits designated at 40 and 41. The alarm circuits include provisions for locking out automatic control of the mill screws and returning the mill to manual control.

The input gage head G1 in position 3-5 is electrically connected to the servo-follower 42 which is a device for sensing thickness of material at position 35 and converting this thickness into mechanical shaft rotation at 43. This shaft rotation which is proportional to thickness operates -a shaft position digitizer 44 with a binary output. This binary output is then fed electrically to the decode and logic unit 45 which serves to eliminate ambiguity to decide which gage (35 or 36) information is to be used, and then in turn feed this information into the magnetic core buffer 46, which then feeds this information into the memory assembly 47. The magnetic core buffer 46 serves to put the input gage information into a form which the memory assembly 47 can accept. This memory is needed because it is necessary to measure the thickness of the material 4 in advance of the bite of lthe mill. It therefore becomes apparent that -the thickness information must be stored, and in effect transported with the ow of the material to the bite of the mill, so that the gage thickness leading the memory corresponds to the thickness of the material at the bite of the mill. The memory 47 comprises, in the example herein explained, an eleven binary bit device having ten cells to progressively store and advance gage readings of the material 4 for every six inches of length thereof. T he eleven binary bits are required to gain reasonable decimal capacity, which in this application is 2,047.

The first nine cells of the memory are of magnetic storage arrangement where the tenth cell is transistorized. Information is read out in serial fashion from this tenth cell to the subtractor circuit represented at 48 and information returned to the tenth cell as indicated by the return path 49. A command circuit 50 extends from the subtractor 48 to the tenth cell of the magnetic memory for synchronizing the operation of this cell so that read-out occurs at the proper instant. Command circuit 50 delivers a pulse from the subtractor 61 when subtractor 61 is ready to perform a subtraction to read out the contents of the last register in the matrix 47 to the subtractor 61.

Since, as pointed out earlier, the gage information is measured before the bite of the mill and then introduced into a memory assembly, it is necessary that the gage information be advanced through the memory in direct relationship with the passage of the strip 4 through the mill. This function is performed by employing the pulses generated in the strip length-sensing unit 13 which is a pulse generator which gives an electrical measurement of the length of material L1 entering the mill in our example. These pulses are counted on the shift preset counter 51 which serves'rto advance the memory assembly 47 each time the pulses accumulated in this counter 51 reachthe preset value. For example, if the input gage G1 is at position 35 and is sixty inches from the bite of the mill, and the memory assembly is ten cellsv deep it becomes necessary to divide the total pulses which would represent sixty inches into tenA equal parts. Thus, if sixty inches ofstrip represented 3,000 pulses, the' shift preset counter v51 would have been set at 300. It is in this manner that the input gage information is correlated with' the strip product. As the shift preset counter 51 commands the memory assembly 47 to advance, it is necessary, for proper operation and synchronization, to interpose the magnetic core drive amplier 52 with the outputthereof connected toV magnetic core buffer 46 to the inputY of theV memory 47. Register shift hold circuit 53 from subtractor 48 to magnetic core drive ampiler 52. delaysY thepulse output therefrom in the event that a command' to shift the memory assembly has occurred while a subtraction was taking place. Upon completion ofthe subtractionthe holdacircuit releases, permitting the magnetic core drive amplifier 52 to perform its function.

The circuit 53` therefore allows a binary number to be completely transferred` from the last cellV of the magnetic memory to the subtractor 48 before a new binary numberis-allowed to enterthe last celll from the precedingcells.l This overcomes the'diiculty that a portion of a numberbeing read-out in serial fashion from the last cell may be lostvinT part orin whole.

When' a measuredgage` reading reaches the tenth cell off the magnetic memory 47', the tenth cell of the memory assembly hasthe inputV gage informationof the in`- crement' of material aboutto enterY the bite of t-he mill. At-this-time-a calculationis made to determine what the input gage should bev toproduce the desired preset output gage of the mill.

As heretofore explained, the'principleinvolved in this invention is set forth byv the followingequation:

where: Y

71=calculated` input gage. l

L2=outputlength- Y L1=input length.

G2=presetdesiredoutput gage.

W=w`idtli factor. u l A K=piclup factor which is a constant derived from the difference in diameter'of the pulse' generator pick-up wheels vwhich ride on idler rolls 1`6and 22.

G1=measuredrinput gage.

The input L1 and output L2 strip length sensing means are at positions 13 and 21 coupled to idler rolls 16 and 22. These pick-ups sample on a continuous basis the incremental length of the material entering the mill and leaving the mill by modulating an R.F. signal in direct relation to the incremental length. To utilize this information in this control system it is necessary to demodulate the signal in the demodulator unit 14. The respective outputs of this demodulator are a series of pulses; the number of pulses on each output line being exactly related to the number of revolutions of the pulse generators 13 and 21 and thus to the strip length. These outputs are fed into the switch unit 54 which establishes according to mill direction infomation receive from mill drive direction selector 55 over circuit 56) which pulse gernerator output should be used as input length rellection L1, as well as which of the two should be used as exit length reection L2. As mentioned in the component description, the outputs of the pulse generators 13 and 21 per increment of stu'p length can be multiplied with multiplier 57 to effect ideal control. The pulses generated by the mill entrance side pick-up L1 are then fed '18 through dual gate 58 to the L1 binary counter 59, while the pulses generated by mill exit pick-up L2are fed through gate 58 and to the L2 binary counter 60.

As referred to in the previous equation, it is necessary to determine the product of KXWXG2. Since these factors Vare all considered constants on any given pass on the mill, they are introduced by rotary switches or pushbuttons in the assembly 6,1, 62, or 63 as the occasion may be. The K factor which represents the relationship between the idler rolls k 16 and 22 andthe pick-up wheels 13 and 21, once established, is not changed; however, it `should be understood that actually there are two distinct values of K and the value used is determined by mill direction (reflected fromv unit 55 over wire 64). The K factors are manually introduced into the assembly 61; the W factor is manually introduced into the assembly 62; and the G2 desired outputgagepfactor is manually introduced into the assembly 63; To satisfy the equation in which these factors are used the factors mustbe v multiplied by each other. This'uis performed atV start-up (when-pushbutton 65` is operated) by multiplying first the K factor' 61 by the W factor 62-, and the resultant product then multiplied by G2 as represented at 63. The nal product is then stored at 66 and' expressed as G2WK; This product then provides the preset limit for the L1 counter which means thatl upon the accumulation in the L1. counter of this preset value, counter 59 will operate the dual gate circuit 58 (through the lead 67) which will block' both the L1 pulses andthe L2 pulses. In addition,a command to read-rout of L2 counter 60 is provided. Upon completion of the read-out into unit 68, both the L1 and the L2 counters are reset to zero, the dual gate 58 is re-opened (via circuit 69, gate 70, and circuit 93), andV theprocess is repeated.

Read-out 68 now contains iny binary form the solution of the desired input gage which the mill should have in order to produce the desired output gage G2. TheV output of readfout 68 provides the calculated value '1 which is applied through lead 71 to the subtractor 48 uponicommand thereof through' lead 94.V The actual measured gage G1 is supplied through lead 72 from the output of the tenth cell of memory 47 to the subtractor 48; The difference between the calculated value and thev actual value of G1- Vas expressed in the foregoing equation constitutes the difference signal which is fed through the unit 73- (which we'rdesignate as theV dead-zone and alarm set) to the unit 74 (designated as time-control .Ol to .l and .l to 1.0 second)r to the mill control 39` heretofore explained.

In order to start this automatic control system, the control'button 65;l of the start unitV 75 is closed. This action provides for the following operation:

(ll) Oscillator 76 is functioned to deliver pulses to the single preset counter unit 77 and to the gate circuit 78. The gate 78 in turn passes the pulses to the magnetic core drive amplifier 52 which serves through the or gate 79 to advance memory assembly 47, such as to clear the memory assembly of prior G1 information and to introduce into each cell the new value of G1. Upon completion of the above, preset counter 77 operates to close the gate 78 and hence disable the circuit.

(2) A pulse is transmitted over lead 80 from the start unit 75 to the Width factor W unit 62 which initiates the multiplication of KX WX G2 -through the unit 81 (WK-l) Upon solution of this product a pulse is provided from the G2WK unit 66 through lead 82 to the and circuit 83.

(3) The pulse on lead 82 when simultaneously appearing at and gate 83 with the pulse through lead 84 (initiated from the start unit 75 when control button 65 is closed) causes a pulse to be passed from and gate 83 to gate and thence to the dual gate 58. Gate 58 is forced to open by this pulse thus starting initially the L1 and L2 counters.

Oscillator 76 is disabled as soon -as counter 77 has 19 reached its preset limit; thus signifying that the memory assembly 47 has been cleared of past information and refilled with new G1 information (all ten calls).

We have constructed, adapted and tested the automatic control system of our invention on a cold rolling steel mill and have found the system of our invention to be very ecient, economical, practical, reliable and precise.

The applicants realize that there are many ways in which the principle of this invention may be carried out and the examples shown herein are to be considered in the illustrative sense and not in the limiting sense.

While we have described our invention in certain preferred embodiments, we realize that modifications may be made therein as well as in 'its application, as for example, instead of controlling screw action on a mill, the system may be utilized for controlling tension of the strip v or a combination of tension and screw action, and we therefore desire that it be understood that no` limitations upon our invention are intended other than may be imposed by the scope of the appended claims.

What we claim as new and desire to secure'by Letters Patent of the UnitedStates is as follows:

l. In a system for performing work upon a moving strip Vby a pair of rolls through which said strip is advanced, means positioned along the path of said strip for measuring the gage of the strip before said strip enters said rolls and for producing data representative thereof, means for measuring the length of strip passing into said rolls and for producing data representative thereof, means for producing a data representation of the desired gage of said strip leaving said rolls, means for measuring the length of strip passing from said rolls and for producing data representative thereof, means for storing data representative of the gage of said strip before entering said rolls measured by the said measuring means, means for combining said data in respect to gage and length of strip before said strip enters said rolls to produce a rst combination, means for combining said data in respect to length of strip after leaving said` rolls with data representative of said desired gage to produce a second combination, means for comparing said first and said second combinations, and means controlled by said comparing means for adjusting said rolls.

2. In a system for performingwork upon a moving strip by a pair of rolls through which said strip is advanced, means positionedY along the path of said strip for measuring the gage of the strip before said strip enters said rolls and for producing data representative thereof, means for measuring the length of strip passing into said rolls and for producing data representative thereof, means for producing a data representation of the desired gage of said strip leaving said rolls, means for measuring the length of strip passing from said rolls and for producing data representative thereof, means for storing data representative of the gage of said strip before entering said rolls measured by said measuring means, means for adding data representative of the initial Width of said strip to data relating length of said strip in said storage means, means for combining said data in respect to gage and length of strip before said strip enters said rolls to produce a first combination, means for combining said data in respectto length of strip after leaving said rolls with data representative of said desired gage to produce a second combination, means for comparing said rst and said second combinations, and means controlled by said comparing means for adjusting said rolls.

3. In a rolling mill control system for operating upon a strip of ductile material by a pair of rolls through which said strip is advanced, a thickness gage for measuring the gage of the strip entering said rolls for producing a single representative thereof, means for converting said signals produced by said thickness gage into entering gage data having a digital notation, a first counter for storing said entering gage data, a second counter capable of receiving data in digital notation representative of predetermined thickness for the exit strip from said rolls, a tachometer pulse generator coacting with the'strip entering said rolls, a tachometer pulse generator coacting with said exit strip, an analog to digital converter for Veach said generator, a gate controlled by the rst said generator to bring about synchronous operation of the second said generator, a iirst multiplier for combining data from first said counter and rst said generator, a second multiplier for combining data from second said counter and second said generator, means for adjusting said rolls, and a differential amplifier vhaving an input fromV each said multiplier for controlling the said adjusting means.

References Cited in the tile of this patent UNITED STATES PATENTS

Images