US3448253A

US3448253A - "quasi-absolute" digital control system for winding filament on mandrel

Info

Publication number: US3448253A
Application number: US325621A
Authority: US
Inventors: Ernest K Bramblett; Richard W Clapp
Original assignee: North American Rockwell Corp
Current assignee: Boeing North American Inc
Priority date: 1963-11-22
Filing date: 1963-11-22
Publication date: 1969-06-03
Anticipated expiration: 1986-06-03

Description

June 3, 1969 E. K. BRAMBLETT n. ETAL 3,448,253

"QUASI-"ABSOLUTE" DIGITAL CONTROL SYSTEM FOR WINDING FILAMENT ON MANDREL Fil ed Nov. 22. 1963 Sheet of 4 INVENTORS ERNEST K. BRAMBLETTII RICHARD W. CLAPP AGENT E. K. BRAMBLETT u. ETAL 3, 8, 53- "QUASI-ABSOLUTE" DIGITAL CONTROL SYSTEM FOR WINDING FILAMENT ON MANDREL Sheet 2 of 4 June 3, 1969 Filed Nov. 22,1963

AGENT E Nm Nm. |||l||||| ulllllll III mm June 3, 1969 E. K. BRAMBLETT n. ETAL 3,

"QUASI-ABSOLUTE" DIGITAL CONTROL SYSTEM FOR WINDING FILAMENT ON MANDREL Filed Nov. 22, 1965 Sheet 3 of 4 2 K 2 l5 8 l 3 LLI I B 11 o I r o 5 l0 I5 MANDREL- ENCODER PULSES FIG. 3

Z 9 E 20 U 3 3 l5 [LI 5 w u l0 5 m :5 2o 25 3o MANDREL- ENCODER PULSES FIG.4

INVENTORS ERNEST K. BRAMBLETTII RICHARD w. CLAPP AGENT June 3, 1969 Filed Nov. 22, 1963 E. K. BRAMBLETT II, ETAL "QUASI-ABSOLUTE" DIGITAL CONTROL SYSTEM FOR WINDING FILAMENT ON MANDREL Sheet 4 014 INCRE MENTOR DO NOT CHANGE NUMBER IN HIGH ORDER REGISTER POSITION INCREMENTOR INCREASE NUMBER IN HIGH ORDER REGISTER POSITION YES INCREMENTOR DECREASE NUMBER HIGH ORDER REGISTER POSITION COMPARATOR IS THE SIGN THE SAME 72 INCREMENTOR DO NOT CHANGE NUMBER IN HIGH ORDER REGISTER POSITION FROM 86L.

YES

I I I I I I I I I l I I COMPARATOR I IS THE NEW l NUMBER LARGER THAN THE OLD NUMBER 5 INVENTORS ERNEST K. BRAMBLETTII RICHARD W. CLAPP AGENT United States Patent 3,448,253 QUASI-ABSOLUTE DIGITAL CONTROL SYSTEM FOR WINDING FILAMENT 0N MANDREL Ernest K. Bramblett II, Canoga Park, and Richard W. Clapp, Tarzana, Calif., assignors to North American Rockwell Corporation, a corporation of Delaware Filed Nov. 22, 1963, Ser. No. 325,621 Int. Cl. G06f /46, 7/48 US. Cl. 235151.1 12 Claims ABSTRACT OF THE DISCLOSURE This invention relates to a digital control system; and more particularly to a system wherein an informationbearing, or program tape controls a machine having a movable feed-head.

Background For convenience, the present invention will be described in terms of the movable feed-head being used for positioning the filament during a filament-winding operation (to be described subsequently); but this particular usage is not to be construed as a limitation, since the feed-head could alternatively hold a shaping or cutting tool, or a similar device.

Filament-winding is a relatively recent innovation, and comprises winding a continuous filament in a cocoon-like pattern to provide a low-weight high-strength shell.

For example, a continuous filament may be wound around the exterior surface of a cylindrical tank in such a manner that several layers of the filament cover each portion of the tank; the filament and the tank combining their strengths to form a much-stronger composite structure. Alternatively, the filament may be wound around unusual-shaped structures, like a nozzle of a rocket engine, to increase the strength of the structure. Another usage of filament-winding is to wrap a structure of basic material that does not have a high strength; cover the wrapping with an additional layer of the basic material; and add still other alternate wrappings and layers of material to form a unitary structure of high strength. Still another usage of filament-wrapping is to form a cocoon like wrapping around a mandrel, and to then remove the mandrel to leave a Wrapped shell.

Generally, the filament is a continuous strand of highstrength mate-rial, such as glass, that is coated with a suitable plastic. When the cocoon-like wrapping is exposed to suitable curing conditions of temperature, pressure, time, etc., the plastic coating of each strand combines with the plastic coating of adjacent strands to form an impervious rfilament-andplastic structure, whose weightto-strength ratio is about one-third the ratio of a similar steel structure. This low weight-to-strength ratio is particularly advantageous in rocket and satellite projects; and is also useful in structures for tank trucks, pressure tanks, silos, oil storage tanks, chemical storage tanks, and similar structures.

An additional advantage of a filament-wrapped structure is that, should the structure be punctured or slashed,

"ice

it can be readily repaired by the use of a fiberglass-plasticimpregnated patch.

Filament-wound structures presently range up to forty feet in length, and fifteen feet in diameter. The large size of these filament-wrapped structures introduces several problems.

First of all, the winding must be extremely accurate, because cumulative errors will place the final windings so far out of position that the overall structure does not achieve the design strength expected from the winding pattern. Secondly, presently-available coating-plastics tend to cure at room temperatures; and this tendency requires that the filament-winding operations be completed as quickly as possible, preferably in a few hours; otherwise subsequent layers of the filament winding are not suitably bonded to the earlier layers. Therefore, the filament-winding machine must be capable of high-speed operation.

Another problem arises in end-winding the [filament around the tops and bottoms of the tanks or mandrels. The curvature of the end-portions requires careful positioning of the filament; otherwise the filament tends to slip off, or else fails to add the design strength to the structure.

Still another problem results from the fact that the winding must be tight; thus requiring a high, constant tension of the filament during winding.

A further problem arises for the fact that, ordinarily, thousands of feet of program tape would be required to control the movement of the feed-head during the severalhour winding interval.

The above problems, and others, dictate that the filament-winding machine have a feed-head that is capable of extremely-fast precise longitudinal and in-out (radial) movements. In addition, the machine must be massive enough to apply a constant fixed tension to the filament. The machine is thus subjected to rapid acceleration and deceleration of heavy masses; these rapid movements and the inertia of the moving parts complicating the precise positioning of the feed-head.

Objects and drawings It is therefore the principal object of the invention to provide an improved filament-winding machine.

The attainment of this object and others will be realized from the following specification, taken in conjunction with the drawings, of which FIGUE 1 shows a pictorial view of a filament-winding operation;

FIGURE 2 shows a block diagram of the control circuitry;

FIGURE 3 shows the uniform feed-head movement produced by the equipment;

FIGURE 4 shows the effect of non-acceptable program information; and

FIGURE 5 shows a block diagram of a specific portion of the control circuitry.

Description of the invention The invention concept Will be understood from FIG- URE 1. Here reference character 10 indicates a tank, mandrel, or other workpiece that is being filament-wound. One or more spools 12, containing plastic-coated glass filament, permit the filament 14 to be fed over suitable guide rollers 16, and through a feed-head 18 mounted on a movable carriage 19, to be wound onto the rotating mandrel 10. The rotation of the mandrel draws the filament from the spool; certain of the rollers, such as 20, acting as tensioning devices to assure that the filament is properly tensioned as it is wound onto the mandrel.

In a single-filament winding operation, only one spool of filament is used; whereas a multiple-filament winding operation draws filaments from several spools, and juxtaposes the multiple filaments to form a ribbon.

The winding is performed as follows. As the mandrel rotates, interconnecting apparatus causes the feed-head positioning-carriage 19 to move along a trackway 24 parallel to the longitudinal axis of the mandrel. A precise positioning-rack 26, located adjacent to the trackway, has its teeth engaged by precision gears mounted on carriage 19; and suitable motors rotate the precision-gears to cause carriage 19 to move to sequential longitudinal positions.

When necessary, other motors, mounted on carriage 19, cause feed-head 18 to move radially, that is, toward or away-from the mandrel 10.

FIGURE 1 shows an end of the mandrel having, for illustrative purposes, two different types of windings.

Reference character 34 indicates a so-called helical winding, which is formed by rotating the mandrel and feeding the filament from a longitudinally-moving feedhead. The space between adjacent filaments, and the slant of the filament is a function of the rotational-rate of the mandrel and the movement of the filament-feed-head 18. The helical winding can thus be closely-spaced or widelyspaced, and may be slightly or steeply slanted, by controlling the movement of the feed-head.

In order to form several overlapping layers of filament, the feed-head passes back and forth along the length of the mandrel, each pass forming an additional layer of filament.

Reference character 40 indicates a polar type of winding. In order to form this type of winding, the mandrel is held substantially still, while the feed-head 18 moves longitudinally from one end of the mandrel to the other, to place a longitudinal Winding along a cylindrical element of the mandrel. As the feed-head moves past the end of the cylindrical-portion of the mandrel, and reaches the hemi-spherical portion of the mandrel, the rotation of the mandrel is accelerated, and the feed-head moves radially inwardly to form the end-portion of the polar winding. By the time that the feed-head has approached the center of the hemispherical-like end of the mandrel, the mandrel has rotated approximately a quarter of a revolution; and the feed-head then backs out to form the rest of the end-portion of the polar Winding as the mandrel rotates another quarter revolution.

When the single-wrap of the end-portion is completed, the mandrel rotation is stopped, and the feed-head again moves longitudinally to the other end of the mandrel, in order to form the polar winding along a cylindrical section of the mandrel diametrically-opposite the first half of the polar winding.

The end-winding process is repeated at the other end of the mandrel.

The total mandrel rotation for this complete polar-wrap is nearly, but not exactly, a full revolution. Thus, the next polar winding is adjacent to the first; and upon continuation of this process, a completely wrapped structure is formed.

It may thus be seen that for this intricate type of winding, the feed-head must race longitudinally the entire length of the mandrel, preferably at a rate of about five feet per second; must then stop its longitudinal motion and perform a radial movement, at a rate of about one foot per second, toward the center of the mandrel; must then back out; and must then race longitudinally to the other end of the mandrelwhereupon the end Winding is again formed.

Other winding-patterns require other types of feed-head movements. It will be understood that the high speed, tremendous acceleration and deceleration of the heavy framework, and the precise positioning requirements impose an unduly heavy burden upon the feed-head positioning and control apparatus.

The filament-winding machine comprises several cooperating assemblies. One of these comprises apparatus for controlling the rotational speed of the mandrel, and for indicating the angular position of the mandrel at all times.

A second assembly comprises a tape-reader that reads the information on the tape; and commands the movement of the feed-head in accordance with this information.

A third assembly comprises apparatus for moving the feed-head in a longitudinal direction, and in a radial direction.

Other assemblies comprise arrangements for sensing the actual position of the feed-head; for determining whether this is the desired position; and, if not, for re-positioning the feed-head to the desired position.

These assemblies are indicated in the block diagram of FIGURE 2. Here reference character indicates an arrangement for controlling the mandrels rotational speed.

This arrangement comprises an adjustable mandrelspeed control 52, such as a potentiometer, whose output may be amplified by a summing amplifier 54, and then applied to a motor 58. The output of the motor is sensed and fed back, by means such as a tachometer feedback circuit 60, to the summing amplifier 54. Thus, if the motorand therefore the mandrel-should happen to accelerate undesirably, the feedback signal causes the summing amplifier 54 to slow the motor by means of a motor speed-control device 56, which could mean closing a valve if the motor 58 is of the hydraulic type. Conversely if the motor should decelerate undesirably, the feedback signal and the summing amplifier act to open the valve, so that the motor is accelerated to reach the desired speed set by mandrel-speed-control 52.

It may be desirable to have the mandrel rotate rapidly during the end-winding operation, and to have the mandrel rotate slowly during the rest of the polar winding or during the entire helical winding. The above-described mandrel speed-control arrangement permits the mandrel speed to be controlled by adjusting control 52 manually. Alternatively, the mandrel speed may be controlled by means such as a cam, whose instantaneous position is a function of longitudinal and/or radial location of the feed-head. In either case, the mandrel speed is then maintained by the above-described arrangement 50.

Reference character 62 indicates a device for detecting mandrel rotation, and thus indicating the instantaneous angular position of the mandrel with respect to an initial starting position.

The mandrel-p0sition-indicating device 62, frequently called an encoder, is of the type that produces a pulse for each given amount of rotation of the mandrel. For example, after the filament-winding operation is initiated, the mandrel-encoder 62 produces a pulse every time that the mandrel rotates a given amount; say every time the mandrel rotates one-thousandth of a revolution.

In actuality, each mandrel-encoder pulse comprises two or more pulses that have a given phase-relation; that is, one of them occurs before the other. If the mandrel were to rotate in the opposite direction, the different phaserelation would indicate backward rotation. The phaserelation and number of pulses is thus used to indicate the position of the mandrel at all times.

These mandrel-encoder pulses are applied to a readout circuit 64 that converts the pulse into a form that may be used by subsequent circuitry. The converted pulses are then applied to a pulse-counting circuit 66 that produces an activating pulse on the occurrence of e.g., every sixteenth mandrel-encoder pulse; the reason for this to be explained later.

As previously indicated, the desired movements of the feed-head have been carefully calculated, and the information has been recorded on a program tape 68 in the form of magnetism or punched holes. For convenience of explanation, the use of a punched-hole tape will be assumed.

In some cases the program tape 68 has information on as many as eight longitudinal channels; and four transverse lines of information may be necessary to define the next location of the feed-head.

It is of course essential that the movement of the tape, and the location of the feed-head be closely and precisely correlated with the rotation of the mandrel.

Rather than trying to maintain independent absolutelyprecise rates for both the rotation of the mandrel and the longitudinal movements of the tape--which is diflicult, or programming both feed-head movement and mandrel-rotational on the tape-which requires prohibitively large amounts of tape, the present invention uses the output of the mandrel-encoder for advancing the tape, and thus commanding the movement of the feedhead to accomplish the desired winding pattern.

This is accomplished as follows:

The activating pulses from the pulse-counting circuit 66 are applied to tape-reader 70, which uses the activating pulses to advance the tape. For example, if the mandrel has rotated sixteen-thousan'dth's of a revolution, the mandrel-encoder 62 has therefore produced sixteen pulses. The sixteenth pulse causes the pulse-counting circuit 66 to produce an activating pulse, which is applied to the tape-reader 70; which thereupon advances the tape in order that the next-sequential batch of information there on can be read.

Thus, if the mandrel should be accelerated-intentionally or unintentionally-the activating-pulses are produced at a faster rate-so that the programmed information on the tape is provided fast enough for the speeding mandrel. If, on the other hand, the mandrel should be decelerated, the activating pulses are produced at a slower rate, so that the programmed information is not provided too fast. This arrangement assures that the tape is read at a rate corresponding with the rotational rate of the mandrel; and obviates the difliculty and the prohibitively large amounts of tape associated with the previously-described approaches.

The tape reader 70 has several electrical outputs. One of these contains information related to the new desired longitudinal location of the feed-head; and this output is applied to a longitudinal-location readout-circuit, 72, where the taped information is converted into a suitable electrical form; and is stored. Here it is checked for error by any of the well-known parity check-circuits 74.

The tape-reader 70 also has an output that contains information related to the new desired radial location of the feed-head; and this output is applied to a radial-location readout circuit 76. Here it is converted into a suitable electrical form; and is stored. The radial-location information stored here is also checked for error by parity check-circuit 74.

Simultaneously, a check is also made to assure that the tape has actually advanced the required number of lines.

Should any of these checks indicate an error, a visual display (not shown) is produced.

A slight digression is necessary at this point.

The tape may be programmed in either of two ways. The first, and simplest--known as the incremental method-programs the feed-head to move a given number of units (increments) from its last location.

For example, if the feed-head is to move from location 32 to location 36, a tape programmed in the incremental manner would command a movement of +4 units (3632); the plus sign indicating movement in the forward direction. Similarly, if the feed-head is to move from location 32 to location 41, the tape would command a movement of +9 units (41-32). In a similar manner, a feed-head movement from location 3-2 to location 28 would require the program tape to command a feed-head movement of -4 units, the negative sign indicating backward movement.

Unfortunately, if the circuitry were to make an error, and produce an incorrect movement, all of the subsequent feed-head locations would also be in error. Moreover, the errors are cumulative; and at the end of the winding operation, the feed-head may be far from its desired location.

However, this incremental mode of programming handles only small numbers (+4, +9, -4); and thus requires a minimum amount of space on the program tape, and relatively simply circuitry.

The second way of programming the tape is known as the absolute method, and commands the feed-head to move to the absolute, or desired, feed-head location.

For example, if the feed-head is to move from location 32 to location 36, a tape programmed in the absolute manner would command a movement to location 36. Similarly, if the feed-head is to move from location 32 to location 41-or to location 28-the program tape commands a movement to the new location.

It will be noted that if the circuitry were to make an error, and produce an incorrect location of the feed-head, the next command of a tape programmed in the absolute manner would direct the feed-head to the next correct location. Moreover, the errors are not cumulative.

It will be noted however, that the absolute mode of programming handles large numbers (36, 41, 28); that is, a representation of the entire new feed-head-location must be stored on the tape. This requires a large amount of space on the tape, and relatively complex large-number circuitry.

The absolute mode of operation requires that the equipment store each new desired location, subtract the last location of the feed-head from the new desired lo cation, and convert the difference into the necessary feed-head movement. While the absolute mode of programming is more difficult to mechanize, it assures that the feed-head will be at the desired location at all times.

A third factor must be considered. Since high threedecimal-place precision is required, and the mandrel may be forty feet long; a five-digit decimal location-number such as 39.000 may be required to define the feed-head location in the absolute mode of operation. Alternatively a smaller four-digit decimal movement-number such as +4.000 may be sufficient to defiine the feed-hand movement in the incremental mode of operation.

As mandrels become progressively larger, the required absolute-mode location-numbers become inconveniently large, and require such a large space on the program tape as to make the program tape irnpracticably long.

Still another factor must be taken into consideration. In view of the required high-precision and smooth operation, the feed-head must undergo a large number of small movements. If all of these movements are programmed, in either the incremental or the absolute system, the tape is again prohibitively long.

As will be seen later, the present invention solves these problems in a novel manner.

Continuing now with the explanation of the windingmachines operation, if the tape-advance check or any parity-check indicates an error, the erroneous feed-head information is ignored; and the subsequent feed-head information from the next programmed batch of information is used. This prevents the feed-head from being moved to an undesired location. Moreover, if the indicated feed-head information indicates an unusual type or amount of movement, the indicated feed-head information is also ignored in favor of the subsequent feedhead informationthe ignoring process and the use of the next batch of feed-head information to be explained subsequently.

Thus, only acceptable feed-head information is stored in the longitudinal and radial position readout-circuits, 72 and 76; and this acceptance and storing operation takes place on the occurrance of every activating pulse.

Occurrence of the next (sec-0nd) activating pulse from pulse-counting circuit 66 causes the tape to advance; and causes new feed-head information to be applied to location-readout circuits 72 and 76, where it is checked and stored as described above.

For ease of explanation, the rest of the explanation will be in terms of longitudinal feed-head location only; it being understood that radial information is processed in the same way.

In a manner that will be understood from the following explanation, the previous desired longitudinal-location of the feed-head has been stored in the absolute longitudinal-location circuit 80L. Similarly, the previous desired radial-location of the feed-head has been stored in the absolute radial-location circuit 84R, the L and R indicating that the circuitry is associated respectively with longitudinal or radial movements of the feed-head. The longitudinal-and radial-

location circuits

84L and 84R then are updated by appropriate circuitry to contain the new desired longitudinaland radial-locations previously read from tape 68 into circuit 72.

In the occurrence of the next activating pulse, register control circuit 84L subtracts the previous feed-head location-as stored in circuit 84Lfrom the new desired feed-head location-as stored in circuit 72; thus obtaining the change in longitudinal feed-head location necessary to place the feed-head in the new desired longitudinal location.

It will be recalled that in order to precisely indicate every location of the feed-head for an interval of several hours, thousands of feet of control-tape would ordinarily be required.

Since it is impracticable to handle such large lengths of tape, the present invention programs the feed-head locations for only every revolution of the mandrel; and the equipment interpolates between these batches of information in the following manner. This procedure shortens the program tape to the length otherwise necessary.

As thus far described, the change of feed-head location for each 7 revolution of the mandrel is stored in register control circuit 80L by the activating pulses that result from every sixteenth pulse from the mandrel-encoder 62; the other intermediate mandrel-encoder pulses being unused as thus far discussed.

In actuality, each of the pulses from the mandrelencoder 62 is used; each set of sixteen pulses dividing (interpolating) the desired-movement into sixteen equal steps. This is done as follows.

Assume that the desired change in feed-head location, as commanded by the tape 68, is stored in the register control circuit 80L; and happens to be sixteen units. Assume furtherfor ease of explanation onlythat the absolute-location circuit 84L is empty, as might be the case at the start of the winding-operation. This latter absolute-location circuit 84L has a register 86L that registers the desired absolute location of the feed-head; the register containing a number of positions. The four rightmost or low-order register-positions 86L are used for interpolation, while the central or high-order registerpositions 86L" accept the overflow from the interpolation register-positions. Thus the absolute-location register becomes filled with desired feed-head location-information. Register 86L also contains a sign position 86L.

By use of the well-known linear-interpolation technique, each of the sixteen pulses from the mandrel-encoder 62 transfers one-sixteenth of the desired position-change information (e.g., of the difference between the old and the new absolute locations) from register control circuit 80L to the lower order register positions 861.. in absolute-position circuit 84L, where the overflow of the newly-transferred information is added to the information already there to provide the location desired for each revolution of the mandrel.

. FIGURE 3 is an illustration of the desired feed-head locations (stored in register 86L) as produced by the linear interpolation technique based on mandrel rotation; the feed-head locations being plotted vertically, while the mandrel-encoder pulses are shown on the horizontal scale. Assume again that the desired change of feed-head location (total vertical movement) is sixteen units. The staircase 71 of FIGURE 3 shows equi-spaced vertical interpolated locations of the feed-head, resulting from the sixteen sequential (horizontally-shown) mandrels-encoder ulses.

p It will be noted that each of the horizontally-shown mandrel-encoder pulses causes the feed-head to move upwards one unit (one-sixteenth of the sixteen-unit desired change). As shown, even if the mandrel-encoder pulses do not occur at regular intervals, as might happen if the mandrel rotation is erratic, the linear-interpolation technique, based on mandrel-encoder pulses, produces uniform one-unit feed-head location changes for each 0 revolution of the mandrel. Thus, without programming the feed-head location for each revolution of the mandrel, the interpolation technique causes the location of the feed-head to change one unit for each onethousandth revolution of the mandrel, regardless of nonuniform mandrel rotation; and without the need for either regularly-timed pulses, constant-speed mandrel rotation, or program-tape control of mandrel rotation.

In this way the program on the tape is reduced to onesixteenth; and the feed head is still very previsely controlled to produce a plurality of small, evenly-spaced movements.

Furthermore, the interpolation overflow into the highorder register positions 86L" of absolute-location circuit 84L causes circuit 84L to be simultaneously and continuously upgraded to indicate the instantaneous desired feedhead location; and its contents may be checked against the programmed information every revolution of the mandrel, if desired, to further assure the accuracy of the operation.

Regressing for a moment, it was previously stated that if the parity check, or some other check indicated an unacceptable feed-head location, this unacceptable location-information was ignored; and the next acceptable feed-head location was used. It may now be seen why this is possible. If such erroneous information is ignored, the register control circuit 80L will not contain the sixteen-unit change in feed-head location between the first and second batches of information on the tape; but instead Will contain the thirty-two-unit change in feed-head location between the first and third batches of information.

The result is shown in FIGURE 4, where the solid-line staircase 73 indicates the intended feed-head location change to be produced by linear interpolation based on mandrel rotation. If however, the first sixteen-unit change information is unacceptable because of a reason such as an unsatisfactory parity check, no feed-head location change is indicated during the first sixteen mandrelencoder pulses; the absence of feed-head movement being indicated by the horizontal portion of dotted-line 75. When the next acceptable programmed information is received, the subtraction circuit indicates a positionchange of thirty-two units; and the linear-interpolation circuit produces sixteen equal position-changes for each of the next sixteen encoder pulses, as indicated by the dotted-line staircase portion of dotted-line 75.

As shown, even though one batch of information was unacceptable, the desired absolute feed-head location is indicated after the thirty-second mandrel-encoder pulse. Therefore, no cumulative errors are introduced.

It will be noted that during the first sixteen pulses of FIGURE 4 there was no change of feed-head location, due to the fact that the information was not acceptable; and that for the second sixteen pulses the location-change of the feed-head was faster than programmed. As a result, the movement of the feed-head, and the winding pattern during these thirty-two pulses is somewhat different than programmed; even though the feed-head eventuates at the desired location.

However, this discrepancy in winding pattern covers a mere of a mandrel rotation; an insignificant deviation that is preferable to an erroneous winding pattern that would have been produced by using the unacceptable information; and is also preferable to stopping the winding process because unacceptable information was presented. Moreover, the previously-mentioned parity warning display calls attention to the presence of unacceptable location information; and the reason for this can be investigated, and corrected.

Thus, the use of linear interpolation shortens the program tape by permitting the programming of only onesixteenth of the feed-head locations,

The present apparatus also uses another tape-shortening technique, as follows. As previously indicated, the absolute-location circuit 84L of FIGURE 2 contains register 86L, which contains a number of register-positions; and is thus able to store the desired feed-head location information in a manner to be described.

As previously indicated, the four right-most register position 86L are used for interpolation; the overflow from the interpolation positions being added to the central high-order register-positions 86L; the left-most register position 86L stores a sign bit.

It will be recalled that if the feed-head location is to change from +32.000 to +'36-.000, the absolute-position mode of programming would require sufficient space on the program tape to indicate the +32.000 and the +36.000 values. These values would require sixteen positions for binary-code storage. Thus, as previously indicated, the absolute mode of programming would require impracticable lengths of tape.

The present invention uses a different concept. The desired absolute-location register 86L contains the old absolute location +32.000 in a binary form.

When the new feed-head position +36.000 is commanded, the actual tape input is a ,quasi-absolute +6.000, rather than a true absolute value of +36.000; that is, the new tape input is only a lower order portion of the new absolute location, such lower order portion having a maximum magnitude only suflicient to define the maximum location change for tape reading. Note that the tape input has a number of orders sufficient to encompass such change, but the digits of the tape input define not the change, but a lower order portion of the new absolute location. The low-order register portion 86L has a size just sufficient to handle such new tape input, and the register control circuitry 80L subtracts the old location (+2.000)--as stored in the register-postiton 86Lfrom the new quasi-absolute of input +6.000, to produce a location-change of -+4.000; the plus sign showing that the feed-head is to move forward 4.000 units, from location +32.000 to location +36.000. Attention is directed to the fact that this is in incremental technique.

Suitable location-change circuitry 85L of FIGURE 2, checks whether the new location 6.000) stored in circuit 72 is greater than the +2.000 currently in the loworder register position 86L; and whether the sign is the same. In this case both conditions exist, so the locationchange (+4.000) is added--by the previously-described interpolation techniqueto the value +2.000, to produce a +6.000 in the low-order register-position 86L; the number 3 in the high-order register position 86L, the +6.000 in the low-order and sign register-positions 86L and 86 now representing the new absolute feedhead location +36.000.

Thus, the modified-binary, absolute, and incremental techniques, plus the use of interpolation, minimized the length of tape and register capacity; while providing extremely precise feed-head location control.

If the feed-head has been programmed to go from position +32.000 to +41.000, the actual tape input would have been the quasi-absolute value +1..000, rather than the true absolute value +41.000; and the register control circuitry 80L would have produced an increment +9.000 (+1000 minus +2.000) for the desired change of feed-head location. This 9.000 would be added by means of the interpolation technique to the 2.000 in the low-order register-positions to 86L to produce a 1.000. Location-change circuit 85L checks whether the new location number (+1000) is greater than the +2.000 currently in the low order register-positions 86L, and checks the sign with that contained in register position 86L. In this case the sign is the same indicating a desired movement in the forward direction; but the new location number (+1.000) is smaller than the old location number (2.000), indicating that the forward movement requires that the feed-head go from 32.000 past the 39.000 feed-head location into the 40s. As a result, the 3 in the high-order register-positions 86L is increased to 4; the in register-position 86L the location number 4 in the high order register-positions 86L", and the 1.000 in the low-order register-positions 86L now representing the new feed-head absolute location +41.000.

If the feed-head had been programmed to go backward from position 32.000 to +28.000, the actual tape input would have been the quasi-absolute value --8.000; and the register control circuitry L would have produced an incremental 4.000; which would have been added (subtracted) by the interpolation technique, to the +2.000 in the low-order register-positions 86L to produce a new location number of +8.000 in register-positions 86L to produce a new location number of +8.000 in register-positions 86L. In this case the sign is different, indicating a desired movement in the backward direction; and the new location-number, +8000, is larger than the old number, +2.000; indicating that the backward movement goes from +32.000 past the 30.000 feed-head location into the 20s. As a result, the 3 in the high-order register-positions 86L is reduced to -a 2; the numbers 2 and 8.000 now representing the new feed-head absolute location +28.000.

More-detailed location-change checking circuitry for controlling the change in the high-order register-positions 86L is shown in block-diagram form in FIGURE 5. As previously explained, this circuitry compares the sign and the value of the quasi-absolute programmed information stored in circuit 72 with the information in the absolutelocation circuitry 84L; and determines whether the numeral in the high-order register position 86L" should be decreased, increased, or left as it is.

It may thus be seen that the disclosed apparatus uses a quasi-absolute, incremental, interpolation concept to provide absolute location of the feed-head; and that this concept permits shortening the program tape to practicable lengths, and also reduces the number of required register-positions. Moreover, as the mandrels become increasingly larger, the advantages of this concept over the pure absolute and incremental programming modes become even greater.

Referring again to FIGURE 2, in order to assure that the feed-head is in the desired location, information is obtained from a longitudinal-feed-head location encoder L that is operated by the carriage-moving arrange ment, and provides information about the actual location of the feed-head in the longitudinal direction. Encoder 100L feeds its information through a suitable readout circuit 102L into an actual longitudinal location circuit 104L.

A subtraction circuit 106L continuously compares the actual location of the feed-head-as indicated by actuallocation circuit 104L, with the desired location of the feedhead-as indicated by register 86L in absolute-location circuit 84L; and produces a feed-hand location error-signal that is fed to signal-utilizing means 108L that produces movement of the feed-head. If the feed-head has already moved to the commanded location, no error-signal exists, and the feed-head is permitted to remain at this location.

If however, the feed-head is not actually at the commanded location, the error-signal acts to increase or decrease the movement of the feed-head; so that the corrected movement brings the feed-head to the commanded location.

Advantages It will be realized that the disclosed apparatus has a large number of advantages. First of all, the quasi-absolute technique permits a shortening of the program tape, and requires fewer register-positions. Secondly the use of the quasi-absolute technique obviates the possibility of cumulative errors. Thirdly, only acceptable feed-head location is used; thus assuring that there will not be any gross errors in the feed-head movement pattern. Fourthly, the interpolation technique also shortens the program tape; and provides a large number of small feed-head movements, without oven-loading either the tape or the tape reader. Fifthly, the use of the quasi-absolute incremental concept permits the use of smaller registers. Sixthly, the use of the mandrel-encoder pulses obviates the need for precisely controlling the mandrel rotation, programming mandrel rotation, or providing precisely-timed clock-pulses. Seventhly, the absolute-location circuitry provides the instantaneous absolute feed-head locations, which may therefore be checked against the program tape, if so desired. And finally, the use of precisely-regulated feed-head movements, controlled by error signals, permits rapid feed-head movements with minimal possibility of overshooting the desired location.

Although the invention has been illustrated and described in detail, it is to be clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, the spirit and scope of ths invention being limited only by the terms of the appended claims.

We claim:

1. In combination with a mandrel to be filamentwrapped, and a feed-head for controlling the feed-point of said filament, the combination comprising:

means for positioning said feed-head to a location designated by a number stored in a first register;

a second register adapted to receive quasi-absolute lower-order feed-head-location information from a tape; and

means for altering the lower orders of said number stored in said first register to correspond with said quasi-absolute lower-order feed-head-location information received from said tape.

2. The combination defined in claim 1 wherein said second register is adapted to accept only non-erroneous information from said tape.

3. The combination defined in claim 2 wherein said first and second registers each are adapted to store said feed-head location information in a binary-coded format.

4. In a system having a mandrel to be filament-wrapped, a feed-head for controlling the feed-point of said filament, and means for positioning said feed-head to a location designated by the content of .a register storing feed-headlocation information in a binary-coded-location format, that improvement comprising:

means for ignoring non-acceptable feed-head-information from a tape, and for accepting non-erroneous quasi-absolute lower-order feed-head-location information from said tape;

means for comparing the binary-coded-location information in said register with the non-erroneous quasiabsolute information from said tape to produce valid feed-head-location change information; and

means for adding said valid feed-head location change information to the lower-order positions of said register in a linear-interpolation manner.

5. In combination with a mandrel to be filamentwrapped, means for rotating said mandrel, and a feedhead for controlling the feed-point of said filament, the combination comprising:

means, comprising a mandrel-encoder, for producing pulses indicative of the rotational position of said mandrel;

means for positioning said feed-head, said means comprising a tape having quasi-absolute feed-head-location information thereon, said information comprising a sign indicating the desired direction of motion of said feed-head and a lower-order portion of a number indicating a new desired feed-head location; a tape reader; synchronizing means, responsive to said mandrel encoder means, for synchronizing said tape-reader with said mandrel;

means, comprising a register for storing feed-head location information in a binary-coded-location format, said register further adapted to receive said quasiabsolute feed-head-location information from said tape reader;

means for ignoring unacceptable feed-head-information from said tape-reader, and accepting subsequent acceptable feed-head-information from said tapereader;

means for combining the information in said register with the acceptable quasi-absolute feed-head-location information from said tape reader, said means comprising means for adding, on the occurrence of every pulse from said mandrel-encoder, a portion of the feed-head change data generated from acceptable quasi-absolute information from said tape-reader, thereby producing in said register the desired instantaneous location of said feed-head;

means for sensing the actual instantaneous location of said feed-head;

means for comparing the instantaneous desired actual location of said feed-head to produce an error signal; and

means for moving said feed-head in accordance wi said error signal.

6. In combination with a mandrel to be filamentwrapped, and a feed-head for controlling the feed-point of said filament, the combination comprising:

means for positioning said feed-head, said means comprising a tape having quasi-absolute feed-head-location information thereon, said information comprising a sign indicative of the desired direction of feedhead location change, and a lower-order portion of a number defining the desired feed-head location;

a register adapted to contain a number indicating the desired location of said feed-point;

means for altering the contents of said register in response to quasi-absolute feed-head-location information from said tape, said means comprising a signcomparator and a magnitude comparator, and means for modifying the number in the low-order register positions to correspond to the low-order information read from said tape.

7. A system for changing the absolute value of a number stored in .a register, said system comprising:

means for storing a group of lower orders of a new absolute value of said number and for storing the sense of the commanded change;

means for determining the sense of the difference between said new value and the absolute value stored in said register;

means for changing the lower orders of said register to the lower orders of said new value stored in said means for storing; and

means for controlling higher orders of said register in accordance with the sense of said change and the sense of said difference.

8. The system of claim 7 wherein said means for changing the lower orders includes interpolation means for effecting such change in fractional increments thereof.

9. The system of claim 8 wherein said interpolation means comprises:

subtracting means for determining the magnitude of the difference between said lower orders of said new absolute value and the lower orders of the absolute value stored in said register;

means for dividing said difference into a selected number of increments, and

means for adding each of said increments to the number stored in said register.

10. In combination with a storage register having storage orders for storing a number N subject to repetitive change and that has a predetermined maximum value of L+C orders and a predetermined maximum variation of C orders for each change, where L is of higher order than means for introducing into said register a new absolute value of N comprising;

(1) quasi-absolute means for generating C orders of said new absolute values and the sense of the difference between such new value and the previous stored value in said register;

(2) means for changing the C orders of said previously stored value to the C orders of the new absolute value; and

(3) means responsive to said sense and to relative magnitudes of C orders of said new and said previously stored values for controlling the value of said L orders in said register.

11. In a filament winding machine having a mandrel to be filament-wrapped, the combination comprising:

(A) a source of filament to be wound on said mandrel;

(B) feed-head means for controlling the instantaneous feed-point of said filament from said source onto said mandrel;

(C) positioning means for positioning said feed-head from a location designated by a given number to another location designated by a new number, said feed-head-positioning means comprising a tape means for storing a group of lower orders of a new absolute value of the new number, said positioning means comprising;

(1) first means for storing the absolute value of said given given number;

(2) means for storing the sense of the commanded change and determining the sense of the difference between the lower orders of said new value and the absolute value stored in said first means; (3) means for changing the lower order of said first means to the lower orders of said new value; (4) means for controlling higher orders of said first means in accordance with the sense of said change and the sense of said difference; and (5), means for moving the feed-head to the position identified by the number now in said first means.

12. In a filament-winding apparatus having a filament feed-head positionable to a location defined by a number stored in a register, that improvement comprising:

means for accepting data comprising a sign indicative of the direction of desired feed-head location change and a lower-order portion of .a new location-defining number,

means for replacing the lower-order portion of the number stored in said register with said accepted lower-order, and

means for altering the higher-order portion of the number stored in said register in response to the sign of said data .and the compared magnitudes of the lowerorder portions of said stored number and said data.

References Cited UNITED STATES PATENTS 2,922,940 1/1960 Mergler 235-151 X 2,964,252 12/ 1960 Rosenberg 242-9 3,062,995 11/1962 Raymond et a1. 3,098,995 7/1963 Mundt 235151.11 X 3,148,316 9/ 1964 Herchenroeder. 3,166,104 1/1965 Foley et a1. 242-9 X 3,172,026 3/1965 Schuman 235151.11 X 3,246,125 4/1966 Morgler 235151.11 X

MARTIN P. HARTMAN, Primary Examiner.

I. KESCHNER, Assistant Examiner.

US. Cl. X.R.