The present invention relates generally to fine tuning the operation of a high speed guillotine cutter at the input portion of a high speed inserter system. In such a system, individual sheets are cut from a continuous web of printed paper for use in mass-production of mail pieces.
Inserter systems, such as those applicable for use with the present invention, are typically used by organizations such as banks, insurance companies and utility companies for producing a large volume of specific mailings where the contents of each mail item are directed to a particular addressee. Also, other organizations, such as direct mailers, use inserts for producing a large volume of generic mailings where the contents of each mail item are substantially identical for each addressee. Examples of such inserter systems are the 8 series, 9 series, and APS™ inserter systems available from Pitney Bowes Inc. of Stamford, Conn.
In many respects, the typical inserter system resembles a manufacturing assembly line. Sheets and other raw materials (other sheets, enclosures, and envelopes) enter the inserter system as inputs. Then, a plurality of different modules or workstations in the inserter system work cooperatively to process the sheets until a finished mail piece is produced. The exact configuration of each inserter system depends upon the needs of each particular customer or installation.
Typically, inserter systems prepare mail pieces by gathering collations of documents on a conveyor. The collations are then transported on the conveyor to an insertion station where they are automatically stuffed into envelopes. After being stuffed with the collations, the envelopes are removed from the insertion station for further processing. Such further processing may include automated closing and sealing the envelope flap, weighing the envelope, applying postage to the envelope, and finally sorting and stacking the envelopes.
At the input end of the inserter system, rolls or stacks of continuous printed documents, called a “web,” are fed into the inserter system by a web feeder. The continuous web must be separated into individual document pages. This separation is typically carried out by a web cutter device. In a typical web cutter, a continuous web of material with sprocket holes on both sides of the web is fed from a fanfold stack from web feeder into the web cutter. The web cutter has a tractor with pins or a pair of moving belts with sprockets to move the web toward a guillotine cutting module for cutting the web cross-wise into separate sheets. Perforations are provided on each side of the web so that the sprocket hole sections of the web can be removed from the sheets prior to moving the cut sheets to other components of the mailing inserting system. Downstream of the web cutter, documents can be transported to a right angle turn that may be used to reorient the documents, and/or to meet the inserter user's floor space requirements.
In a typical embodiment of a web cutter, the cutter is comprised of a guillotine blade that chops transverse sections of web into individual sheets. This guillotine arrangement requires that the web be stopped during the cutting process. As a result, the web cutter transports the web in a sharp starting and stopping fashion.
In a feed cycle, the paper is advanced past the blade of the guillotine cutter by a distance equal to the length of the cut sheet and is stopped. In a cut cycle, the blade lowers to shear off the sheet of paper, and then withdraws from the paper. As soon as the blade withdraws from the paper path, the next feed cycle begins. The feed and cut cycles are carried out in such an alternate fashion over the entire operation.
In some web cutters, it is desirable to achieve a cutting rate of 25,000 cuts per hour or more, for example. This means that the web cutter has a feed/cut cycle of 144 ms. Typically the length of the cut sheet is 11 inches (27.94 cm). If the time to complete a cut cycle is about 34 ms, then the total time in a feed cycle is 110 ms. This means that the web must be accelerated from a stop position to a predetermined velocity and then decelerated in order to stop again within 110 ms. As guillotine cutters are required to generate pages even faster (up to 36,000 cuts per hour), precise motion control coordinated over various mechanisms must be implemented in order to eliminate web breakage and to reliably cut sheets of proper length at high rates.
In this environment, it is important to be able to precisely control the guillotine cutter to accurately perform its cuts during the brief time window available. Since the guillotine blade servo motor is subject to varying torques throughout the up and down cycle of the guillotine blade, it has been found to be difficult to tune the driving servo motor in order to achieve the exacting performance required.
For a typical closed loop motion control system with fixed hardware gains and servo update rate, determining servomotor tuning coefficients is a function of inertial and friction loading reflected back to the servo motor. For mechanisms that have inertial and friction loads that are not constant, determination of tuning coefficients that provide satisfactory or optimized motion control performance can be difficult, if not impossible to achieve. One such mechanism that has varying friction and inertial properties reflected to the motor shaft is a crank-rocker mechanism. The crank-rocker mechanism is typically utilized as a means to provide motion to a guillotine cutter blade assembly.
The present invention provides a method for improved tuning of servo motors used to drive guillotine cutters. Rather than providing a single tuning coefficient to the motor, the tuning coefficient is continuously varied during the blade's cutting cycle. The novel method for selecting the varying tuning coefficients allows rapid and precise cutting and minimizes lag or overshooting.
In a first step of the tuning process, a plurality of discrete positions in the blade cycle are selected for analysis of the optimal tuning coefficients at those positions. For each of those discrete positions, tuning coefficients are determined. In one preferred embodiment, the motor is commanded to move through approximately three degrees (of the three hundred sixty blade cycle) at the discrete position. The actual displacement corresponding to the command is observed. The tuning coefficients for that discrete location are then determined by adjusting the coefficients up or down, and repeating the test until the desired motion is achieved. In the preferred embodiment, the step of determining tuning coefficients is done using PID (proportional, integral, derivative) control techniques with a PID controller providing control signals to the motor amplifier.
After the tuning coefficients have been determined for the discrete locations in the blade cycle, the coefficients for the remainder of the blade cycle are determined through interpolation. In a preferred embodiment, linear interpolation is used. The controller then applies the measured and interpolated coefficients to the amplifier that controls the motor.
In the preferred embodiment, the step of selecting the discrete positions includes selecting 90 degrees, 180 degrees, 270 degrees, and 360 degrees in a guillotine blade cycle. These four positions roughly correspond to peaks and valleys in the coefficients needed to work with the varying torques that are required over the blade cycle. The 180 degree position represents a bottom dead center position and 360 degrees represents a top dead center position in the blade cycle. These top and bottom positions also represent points in the cycle with low torque requirements and low tuning coefficients. The horizontal positions of 90 and 270 degrees represent high torque positions that will require peak coefficients.
One of skill in the art will understand that the gearing ratio of the motor to the blade cycle need not be one to one. Thus, more or less than one rotation of the motor can result in one cycle of the blade. The tuning coefficients are based on the blade position, regardless of the gearing ratio between the blade cycle and the motor.
By testing for the proper coefficients at those four discrete quadrant positions, the appropriate bases for linear interpolation are achieved. Interpolation may also be done based on a sinusoidal shaped curve.
Further details of the present invention are provided in the accompanying drawings, detailed description, and claims.
a, 1b, and 1c depict a view of a guillotine cutter blade cutting across a sheet of web in varying stages.
a-1c depict the guillotine cutter 21 through a downward cutting motion, starting at a beginning position in 1a, to a finished cut position in 1c. Guillotine cutter blade 21 preferably has an edge that is vertically inclined at an angle above the path of web 120. As the blade 21 is lowered (
In this example, TDC and BDC have small moment arms and require lower torques for those positions. Friction is also low on the blade 21 at TDC and BDC, which is a further reason for low torque requirements at those positions. Accordingly, it is expected that motor 22 will require less gain to be driven at those positions.
Positions A-H of the rotary motor 22 in
Positions C and F have large moments arms, and therefore greater torque requirements on motor 22. At position C, paper is being cut, adding a further frictional component. At position F, the blade 21 is being raised against the force of gravity, and will thus require a larger torque output from the motor 22. Accordingly, it is expected that larger gains will be needed at positions C and F for tuning the control of the motor 22.
The blade 21 begins its motion profile 63 when the displacement of the cutter transport is such that, after the blade 21 has reached displacement A, the cutter transport will have come to rest. Blade displacement, A, is the blade position from TDC where the blade just contacts the inner sheet of web 120 minus some amount for margin (includes servo settle time).
The use of closed loop position control systems, as illustrated in
The DAC value is scaled accordingly to match the inputs and outputs of the power stage or amplifier 73. Such scaling is achieved with a digital filter that contains tuning coefficients. The filter outputs a percentage of the range between maximum and minimum values that can be applied to the amplifier 73. In addition to providing the proper gain for the system, the tuning coefficients are also selected to provide desired position accuracy, desired system response and stability. The tuning coefficients may also be referred to as the “gain” of the system. The tuning coefficients may also be characterized as a sum of a subset of parameters that contribute to system stability. In a PID system, proportional gain, derivative gain, and integral gain are the primary components for determining the overall gain. These, and other less significant tuning parameters, are well known in the art and need not be described in further detail here.
Many commercially available amplifiers 73 use+/−10 VDC as an acceptable analog input signal. The power stage 73 converts this input signal and outputs a winding current that is proportional to the input signal. With new components, the digital filter 72 may output a digital value whereby the power stage 73 can accept this digital value and accomplish the same as the analog version. Winding current is delivered to the motor 74 and is typically proportional to motor 74 output torque. This ultimately provides motion to the mechanism 76. An encoder 75 or other suitable feedback device located on the motor 74 or on the mechanism 76 provides the actual position back to the summing junction 71, completing the closed loop. In an inserter machine application, this entire process typically updates at a period of 500 microseconds (or 2 KHz), ultimately providing the desired quality of motion at the cutter mechanism 75.
In the preferred embodiment, tuning operations are performed at separate positions in the cutter blade 22 cycle. Tuning is preferably performed at TDC (0 or 360 degrees), position C (90 degrees), BDC (180 degrees) and at position F (270 degrees) as depicted in
The system is then tested again using the new tuning coefficient, and the resulting operation of motor 74 is observed. One of skill in the art will be familiar with tuning processes for adjusting gains to find an optimal tuning coefficient, and further details need not be included here.
In the preferred embodiment, the tuning coefficients are tested and determined in this way for the four quadrant points of the blade cycle (90, 180, 270, and 360 degrees, also shown as positions C, BDC, F, and TDC in
In the preferred embodiment, tuning coefficients for untested points between these tested quadrant points are determined using interpolation. Linear interpolation is appropriate, but curved interpolation algorithms may also be used.
For an example of linear interpolation, lets assume we know the tuning coefficient XTDC for the position TDC and the tuning coefficient XC for the 90 degree position (position C in
X=((XC−XTDC)(θ−θTDC))/(θC−θTDC)
Linear interpolation is an algebraic process that is easily accomplished when the correct parameters are known.
For interpolation to be useful, it is important that the tested data points reflect the high and low points in the range of proper tuning coefficients. For example, if only TDC and BDC were tested, interpolation would be useless, since none of the higher tuning coefficients needed for the higher torque scenarios at 90 and 270 degrees would be recognized. For the preferred embodiment, that is why the four quadrant points were selected for testing, and for the basis of the interpolation.
Although the invention has been described with respect to a preferred embodiment thereof, it will be understood by those skilled in the art that the foregoing and various other changes, omissions and deviations in the form and detail thereof may be made without departing from the scope of this invention.
Number | Name | Date | Kind |
---|---|---|---|
5070465 | Kato et al. | Dec 1991 | A |
5121663 | Tahara et al. | Jun 1992 | A |
5371449 | Tajima et al. | Dec 1994 | A |
5892345 | Olsen | Apr 1999 | A |
5910721 | Hayashi et al. | Jun 1999 | A |
5936366 | Hamamura et al. | Aug 1999 | A |
6037738 | Morita et al. | Mar 2000 | A |
6658370 | Christ et al. | Dec 2003 | B2 |
Number | Date | Country | |
---|---|---|---|
20080141838 A1 | Jun 2008 | US |