Cutter Distance Sensor for an Adaptive Position-/Torque Control in Cross Cutters

Dirk Hansen Kollmorgen Seidel GmbH & Co. KG

Hardware Development Wacholderstrasse 40 – 42

D-40489 Duesseldorf Germany

Prof. Dr.-Ing. Joachim Holtz

Fellow, IEEE Wuppertal University

Electrical Machines and Drives Fuhlrottstrasse 10

D-42097 Wuppertal Germany

Prof. Dr.-Ing. Ralph Kennel

Senior Member, IEEE Wuppertal University

Electrical Machines and Drives Fuhlrottstrasse 10

D-42097 Wuppertal Germany

Abstract - Cross cutters are important equipments in paper industry. A well-known design is to use mechanically coupled cutters providing synchronous motion. To improve dynamic behavior of cross cutters and to reduce maintenance effort it would be advantageous to operate the cutters by separately controlled servo drives. This idea requires very exact knowledge concerning cutter positions to ensure proper operation and to avoid any crash. This paper presents a progressive sensor measuring the distance between the cutter knives and distinguishing between knives approaching or touching each other directly on the one hand and normal cutting operation with material between the knives on the other hand. Measuring results obtained by a test example of the sensor are presented.

Index Terms – Cross Cutter, Cut Size Sheeters, Cutter Layboys, Cutter Knives, Knife Detectors, Distance Sensor, Position control, Torque control, Adaptive Control

I. INTRODUCTION (INCL. STATE OF THE ART)

Cross cutters (cutter knives), are used for processing of endless pasteboards, cardboards and plastic foils. The basic principle of a cross cutter is shown in Figure 1.

In paper industry most cross cutters are synchronous cutters – i.e. cross cutters driven at synchronous speed where the length of the web cutoff is defined by the circumference described by a single rotation of the cross cutter cylinders. These are used in the fine paper industry in machines called "Cut Size Sheeters" and in the pulp industry as "Cutter Layboys". Asynchronous cutters knives (as discussed in this paper) today are mainly used in the corrugating industry and in so-called "Folio Sheeters".

The speed of the web to be cut is controlled by a different machine section - in the case of a corrugating (fluting) machine the rate is set by the corrugating (fluting) operation. The cutoff length of the product has either to follow a variable reference value (cut to length) or to be synchronized to figures

printed on the web material (cut to mark). The cross cut must always be made at the speed of the web - the speed of the cross cut cylinders must be coordinated with the web speed while the knife blades are cutting. The length of the product is determined by the dwell time of the cross cutter when the knife blades are not in contact.

When the cross cutter is operated at the same speed as the moving web the cutoff length will be the synchronous length determined by the circumference described by the knife tips. When the cross cutter is decelerated and then re-accelerated after cutting the cutoff length will be longer than the synchronous length. When the cross cutter is accelerated and then re-decelerated after cutting the cutoff length will be shorter than the synchronous length.

Most cutter knives use a scissors cutting technique. The knife blades actually spiral around the cylinders making the cutting point actually "scissor" or progress across the web for a finite period of time. This results in synchronizing the knife blades to web speed and in torque sharing mode for a significant angle of the full rotation of the cylinders. There are some cutter knives realizing the cut within a single instant of blade contact (instantaneous cutter) but these become fewer and fewer as the mechanical requirements for a scissors type cutter are not as severe as for an instantaneous cutter.

Fig. 1 Basic principle of a cross cutter

This paper presents a cutter knife design for both domains the synchronous and the asynchronous cutters with scissors cutting technique. A well-known design is using mechanically

coupled cutters to provide synchronous motion of cutter cylinders. The cutting cylinders are braced by torsional force. With respect to this mechanical design a lot of control problems are investigated in the past (e. g. published in [1], [2]). The contact force between the knives has to be adjusted by a couple of adjusting devices (e.g. screws). It is necessary to readjust the equipment rather often for maintenance reason. An improvement of this situation would be appreciated by printing and foil production industry.

II. CROSS CUTTERS WITH SEPARATELY CONTROLLED DRIVES

To improve dynamic behavior of cross cutters and to reduce maintenance effort it is advantageous to operate the cutters by separately controlled servo drives. Figure 2 presents cross cutters of that type. When using cutter knifes with separate drive control there is no basic difference any more in cross cutters for synchronous operation and for asynchronous operation. This might encourage paper industry to make more use of the advantages of asynchronous operated cross cutters. Replacing mechanically coupled cutters by separately driven cutters it is necessary to know exactly the position of each cutter during operation. The controls of both servo drives must take into consideration whether the cutters are revolving, approaching or in contact to each other. Furthermore there is an important difference between the operation with or without paper (or different material) between the cutter knives.

synchronous maschine

sensor

knife base

incremental–encoder

cutter cylinder

Fig. 2 Cross cutter

Fig. 3 Evaluation cross cutter

The new proposal is to operate one of the cutters in

position control (angle–time–trajectory according to the reference signal) and the second cutter using adaptive position-/torque control.

As long as the cutters are not in contact to each other, the second cutter (slave) is operated in position control with a position reference adapted to the (real) position of the first cutter drive (master). As soon as the cutter knives close contact, the second servo drive is switched to torque control with a torque reference considering the fact whether there is paper between the cutter knives or not. When opening the gap between the cutter knives the second (slave) servo drive returns to position control again. Figure 5 presents the complete control structure providing the type of operation described above.

This concept provides a couple of advantages:

• No readjustment action required even after a remarkable time of wear

• Better quality of the cutting process • Longer life time and less maintenance effort for the cutter

knives • Simple adjustment with respect to different cutting

materials and different cutting widths A. Test Equipment

The test cross cutter shown in Figure 3 can be compared

with the scheme in Figure 2. Beside the cross cutter itself the complete evaluation system presented in Figure 4 contains a Rapid–Prototyping–System based on a standard personal computer (166–MHz–Pentium–Processor) and an industrial inverter.

Fig. 4 Evaluation equipment

III. CUTTER CONTACT SENSOR

Information concerning the condition of the knives is necessary to optimize the wear of the cross cutters. A cutter contact sensor can provide this information. There are several possibilities to design a cutter contact sensor.

A. Detecting Knife Contact by Measuring the Ohmic Resistor

Usually both knives of a cross cutter are connected to the knife bases which are electrically conductive and which are at ground potential. The contact force is proportional to the ohmic resistance of the knife blades contact point. To measure the resistance one knife on a separate potential is essential. Therefore one or both of the knives must be mounted isolated on the knife base, see Figure 6.

ω

knife blade

isolation

knife base

cutter cylinder

shaft end

Fig. 6 Isolated knife on the knife holder

ϑ

ω* t*

U*a,b,c

I*α

e+jϑ

I*β

ϕfor

ϕact ωact

tfor

Iβact Iαact

ωfor

hardware

sheet length command actual material velocity

model

positioncontroller

speedcontroller

motormodel

currentcontroller

Fig. 5(a) Control structure of the first (master) cutter drive

ϑ

ω* t*

U*a,b,c

I*α

e+jϑ

I*β

ϕact

ϕcor

ϕfor

ϕact ωact tcor

tfor

kc

Iβact Iαact

ωfor

hardware

sheet length command actual material velocity

cutter contact sensor&

logicevaluation

model

ϕforωfor tforϕactωact

Iβactϕcor tcor

ω* t*

kc

position feed forward

speed feed forward torque feed forwardactual position

actual speed

actual currentα−

actual currentβ−

speed command torque command Iαact

position correction torque correction knife contact signal

positioncontroller

speedcontroller

motormodel

currentcontroller

Fig. 5(b) Control structure (incl. distance sensor) of the second (slave) cutter drive with adaptive position-/torque control

Sliding contacts are used for signal transmission. They are

mounted close to the center of the shaft end, (details see below). The knives form a conducting loop as long as they are in contact to each other. The ohmic resistance of the contact point is in the range of milliohms. Consequently the measuring signal is superimposed by parasitic noise. For minimizing the noise a filter is necessary. The cylinder is mounted rotatable on ball bearings. The electric resistance of the ball bearings is not constant and depends on the bearing lubrication and geometrical tolerances. To provide acceptable transmission the cylinder with the knife blade not isolated must be contacted additional via sliding contacts. The contact resistance of the sliding contacts fluctuates in the range of the measuring signal. Hence it is not possible to detect more information than contact or no contact of the knives.

B. Sliding Contact Construction

There are two possibilities to mount sliding contacts on the shaft end of a cutter cylinder. The contacts can be either mounted on the circumference, Figure 7(a) or at the end of the shaft, Figure 7(b).

a) b)

shaft end

sliding contact

Fig. 7 Sliding contact design

The wear of the centered contact is lower. The reason is

the contact force needed at this position being lower. The contact surface in Figure 7(a) is mounted at the periphery of the cylinder. It has small unbalance tolerances. At high speed it leads to fluctuating contact forces and thus to fluctuating contact resistances. The effect can be minimized by using a contact spring with bigger force. But that results in shorter contact wear. Another possibility to reduce the differences of contact variation is to use several contacts allocated around the contact surface.

C. LC–Resonant Circuit

Using an LC–resonant circuit it is possible to detect the contact of the knives without any wear. The inductance of a cutter contact sensor of this type is formed by a inductive transformer. The signal transmission is working similar to a video recorder. The signals taken from rotating video heads are transmitted inductively via a head transformer from the head cylinder to the head pre–amplifier.

primary winding

secondary winding

ferrite core

Fig. 8 Transformer design

The transformer has two ferrite cores with a

semimonocoque design, [4] containing either the primary or the secondary winding, (Figure 8). One core is mounted concentric at the shafts end of the cutter cylinder. The other core is mounted stationary opposite to the first one with an air gap of 1 mm, Figure 9.

synchronous maschine

gear belt

cutter cylinderfixed primary winding

rotating secondary winding

Fig. 9 Mechanical design

If a knife blade is mounted isolated on the cutter cylinder,

as shown in Figure 6, the capacitance CX is formed between the knife blade and the cutter cylinder. Together with the transformer the resonant circuit shown in Figure 10 is formed. There are several ways to stimulate and evaluate the circuit.

Lh Cx

Lh main inductanceCx capacitance

L1σ L2σ

L1σ primary leakage inductanceL2σ secondary leakage inductance

Fig. 10 Resonant circuit

D. Evaluating Impulse Responses

Both cutter cylinders are electrically isolated against the rest of the mechanics. That results in a capacitance C2 between the cutter cylinders. One knife blade is isolated to the cutter cylinder as describe above. Between the knife blade and the cutter cylinder there is a capacitance C1 (Figure 11). It is possible to measure capacitance C1. While the knife blades are in contact to each other during the cutting process both capacitances are in parallel connection.

C1

C2

transformerknife blade

cutter cylinder

Fig. 11 Parallel connection of the capacitances when knives contact each other

To avoid the transfer of any additional supply energy to the

rotating knife blade, the measurement is realized by a LC resonant circuit (Figure 10). The inductive part is formed by the main inductivity of the transformer. The capacitive part is formed by the capacitance of the knife system. When the effective capacitance of a resonant circuit changes the natural frequency changes as well corresponding to equation (1).

LC1

=ω

(1)

The variation of the natural frequency in the measuring device is measured between 475 kHz and 375 kHz. A short current pulse of 0.2 µs duration is impressed each 10 µs to the primary winding of the transformer. By evaluating the decaying oscillation the oscillation period can be measured.

The electrical circuit consists of two parts, an analog and a digital one (Figure 12). The analog circuit generates the current pulse and detects the zero crossing. The digital part processes and evaluates the signals for the signal computer system.

trigger signal

sign ofimpuls excitation

analogcircuit

interface tosignal computer

FPGA

Fig. 12 Evaluation of the impulse response

The digital circuit is realized as a Field-Programmable Gate

Array (FPGA) providing very flexible processing of the Data. The circuit works with a basic cycle of 10 µs. At the beginning a short current pulse is impressed to the oscillator. After 2 µs, when high frequent parasitic oscillations have decayed, three zero crossings of the response signal are evaluated. The measured period has a resolution of 100 ns. Depending on the data in the control register the measured times are added or set to zero. In another register the actual value is stored after processing to be accessed at any time.

The following figures show measuring results obtained by the analog circuit. The knife blades are not in contact in Figure 13 and in contact in Figure 14. The high frequent

parasitic oscillations can be seen clearly as well as the change of oscillation frequency.

t

5V

0V

2V

0V

5V

0V

-5V

1µs/div

c)

a)

b)

Fig. 13 Cutter contact sensor signals, no contact

a) excitation pulse b) response of the LC oscillator c) zero crossing detection signal

t

5V

0V

2V

0V

5V

0V

-5V

1µs/div

c)

a)

b)

Fig. 14 Cutter contact sensor signals, contact

a) excitation pulse b) response of the LC oscillator c) zero crossing detection signal

Figure 15 shows the output data processed by the digital circuit during motion along the whole cutting area. The signals for knife contact and sliding knives are very different to each other. As long as the knives are within the cutting area the “cutting area signalization signal” (upper trace) has the value 2 V, otherwise 0 V.

t

2V

0V

2V

0V

4V

2V

0V

10ms/div

c)

a)

b)

Fig. 15 Cutter contact sensor signals, cutting area

a) cutting area signalization signal b) knife contact c) sliding knives

The knives can be recognized to be not in contact in the

complete cutting area (signal level around 2.9 V). The reason is the controller not works optimal. The signal occasionally shows signal levels about 0.4 V. In this position the distance of the knives is in the range of tenths of millimeters and the natural frequency already changes during to the influence of the couple capacitance CC between both knife blades, (Figure 19). As mentioned before the signal evaluation is realized by measuring the time between three zero crossings of the response signal from the LC oscillator. The resolution is 100 ns and the natural frequency varies between 475 kHz and 375 kHz. The difference between the times at knife contact and none contact is about 1.1 µs (11 counts) and therefore very small. Hence it is not possible to extract an information about the distance of the knives. It is only possible to detect whether the knife blades are in contact or not.

E. Natural Frequency Evaluation

Evaluating of the natural frequency the same design of the cutter cylinder is used as the one for impulse response evaluation. The resonant circuit is stimulated with the natural frequency (460 kHz) appearing when the distance between the knives is about 1 mm and the knives are not in contact. If the knife blades close contact the capacitance C2 is paralleled to C1 (Figure 11). Consequently the resonant circuit is detuned and the signal amplitude decreases. The signal is evaluated by the help of a peak value rectifier and an additional logic. Due to the use of the special transformer, (Figure 8), the cutter contact sensor has no wear. Figure 16 shows the knife contact signal kc from the sensor while the cutter knives are in contact (signal level 2 V). The amplitude decreases when approaching the area in which a contact takes place. That can be explained by the couple capacitance of the second knife. The couple capacitance increases while the distance of the knife blades is decreasing. The capacitance depends on the geometry of the cutter cylinder

and varying along the cutting area. At the beginning of the cutting area the capacitance of the test equipment was CC = 10 pF and increased until to CC = 17 pF. These values are the maximum values of the couple capacitance without any contact of the knives.

0.160.10 0.22 0.28 0.34

10

8

6

4

2

contact signal

rotation angle

kc

Fig. 16 Signal behaviour, contact in the whole cutting area

0.160.10 0.22 0.28 0.34

10

8

6

4

2

contact signal

rotation angle

kc

Fig. 17 Signal behaviour, sliding knives

The cutter contact sensor described above provides good

detection of the knives being in contact. Figure 17 shows the signal while the knives are sliding. The real distance of the knife blades cannot be measured due to the ratio between the capacitance C1 (330 pF) and the couple capacitance CC being too big.

Every 10 µs a test of knife contact takes place, to obtain more information than just the momentary value of the measuring signal. The contact signal is compared with a reference signal of 4 V. If the value is greater the knife blades are not in contact, otherwise they are. The result of the comparison is stored in a register on the evaluating board. The average value mb of the comparison along the cutting area is shown in Figure 18. The value indicating the desired sliding knives is zero. During contact the output signal is “1” and “-1” as long as there is no contact.

0.160.10 0.22 0.28 0.34

sliding knives

no contact

knife contact

rotation angle

kc

Fig. 18 Average value of comparison

F. Couple Capacitance Evaluation

By evaluating the coupling capacitor CC (Figure 19) the capacitance between the knife blades is measured. Figure 20 shows the structure of the evaluating circuit consisting of two printed circuit boards. The data logger board generates the stimulating signal and measures the coupled signal. The evaluation board provides the measured signals for the signal computer system.

Both knives are mounted on the knife base with electrical isolation forming the capacitors C1 and C2 (Figure 19). Capacitor C1 is supplied by a sinusoidal stimulating signal with a frequency of 10 MHz. Capacitor C2 allows measuring the coupled stimulation signal.

C1

C2

stimulating signal

reference potential

CC

measuring signal

Fig. 19 Coupling capacitor

The signals at the marked points in Figure 20 are presented

in Figure 21.

U

Tsample Treset

T

t1

2

3

TTT

= 100 ns= 12,5 ns= 37,5 ns

sample

reset

Fig. 21 Signal evaluation

S&H S&H AD

C2C1

logic

reset

sampletriggerPC

1 2 3

data logger board

evaluation board

Fig. 20 Coupling capacitance evaluation hardware

The amplitude of the coupled signal increases with the capacitance of the coupling capacitor. The capacitance increases while the distance of the knives is decreasing. The capacitance also depends on the geometry of the cutter base. At the beginning of the cutting area the knives do not overlap. The overlapping and therefore the effective area of the coupling capacitor increases to the end of the cutting area. As described above the coupling capacitance varies between 10 pF and 17 pF. The coupled signal at C2 is lead through a band pass filter to a peak value rectifier. The output of the rectifier is detected by a sample and hold unit on the data logger board. The evaluating board digitizes the actual values. Figure 22 shows the measuring signal along the knife distance. At contact the value is “1” and decreases while the distance is increasing. The increase of the amplitude from the beginning to the end of the cutting region can be seen clearly.

0.050 0.1 0.15 0.2rad

1

0.8

0.6

0.4

0.2

kc

contact signal

cutting region

Fig. 22 Measuring results

This cutter contact sensor provides detection of contact and

knife distance. Using a band pass filter becomes the sensor insensitive to parasitic noise disturbances.

G. Conclusions

A cutter distance sensor with good performance is a requirement to introduce separately controlled servo drives in cutter knives. The test results obtained by test samples of the proposed sensor show that all information necessary for adapting the control structure and control parameters can be obtained by the proposed sensor concept. This type of sensor enables further progress in industrial drive applications, which leads to more replacement of mechanical designs by servo drives with flexible control.

H. References

[1] Th. Eutebach, J. M. Pacas, Damping of Torsional Vibrations in High Dynamic Drives, EPE 1999, Lausanne, Switzerland

[2] J. M. Pacas, A. John, Th. Eutebach, Automatic

Identification and Damping of Torsional Vibrations in High-Dynamic-Drives, ISIE 2000, Puebla, Mexico

[3] R. Kennel, Accuracy and Resolution in Servo Drives -

What is the difference ? What is important ? What is the bottle neck ?, EPE - PEMC 2000 (Power Electronics and Motion Control), Košice, Slovakia, Sep. 5-7, 2000

[4] Standard Bauteil Service (German language),

page 57 – 59, EPCOS AG, 2000