Imaging system for processing a media

The invention relates to an imaging system for processing a media, comprising a media transport path, an imaging station arranged along said media transport path, displacement means for controllably displacing the media along the media transport path relative to said imaging station and a controller assembly.

In known imaging systems the media is positioned relative to the imaging station by means of commonly known transport pinches, which are driven by electric motors. The increasing demands for higher image quality and speed result in increasingly strict demands of positioning precision of the media with respect to the imaging station. For example, in a printing system, wherein an image of marking material is applied on a print media, the print media is stepwise displaced relative to the printing station such that the image can be applied in several swaths. In such systems print media have to be positioned at the exact required position when the marking material is applied. Any deviation of the position of the print media relative to the printing station may result in a degraded image quality, as a result of misplacement of particles of marking material on the print media. In general, due to the stricter positioning requirements, it becomes increasingly more difficult to satisfy the strict positioning tolerances. This imposes higher requirements for the mechanical construction of the displacement means of the media and for the specifications of the electrical drive means that are used for driving the displacement means. In general, this leads to an increasingly more complex and expensive construction of the known imaging systems.

It is an object of the invention to provide an imaging system with an increased performance with respect to the positioning of a media, without increasing the complexity of the mechanical structure.

To this end, according to the invention, the controller assembly comprising a feedback filter, a feedforward filter, a low-pass filter and a memory for storing and time delayed releasing control data, where in operation the displacement means are actuated in response to an actuation command generated by the controller assembly, the actuation command having a feedback component based on a filtering by the feedback filter

of an error signal comprising information about the position error between a desired position and the actual position of the media and a feedforward component based on a time delayed, low-pass filtered, frequency dependent filtering of the error signal by the feedforward filter, the feedforward filter being devised such that the closed-loop controlled characteristics of the displacement means are compensated.

Thus, the positioning requirements of a media relative to the imaging station are met or even improved, while the mechanical complexity of the overall imaging system is not increased. The feedback component is used to correct for incidental errors while the feedforward component corrects for structural influences that negatively influence the positioning of the media. Incidental errors may for example include disturbances due to ground vibrations as a result of the operation of neighbouring instruments, or manual disturbances imposed on the media or on the media positioning means. Structural influences may include for example the unroundness of an axle or skew of a driven pinch roller.

In an embodiment the feedforward filter is devised such that the frequency transfer function of the feedforward filter is substantially equal to the inverse of the process sensitivity of the controlled displacement means. As the process sensitivity is a good indication for the behaviour of the closed-loop controlled system, the compensation of the closed-loop controlled system characteristics is well reached by the implementation using the inverse of the process sensitivity.

The better the feedforward filter compensates the closed-loop controlled behaviour the better the feedforward component will be able to improve the performance. The process sensitivity may be theoretically modelled or measured, e.g. by a frequency response measurement. The implementation of the feedforward filter may be adapted to correct for any occurring instabilities, due to unstable poles or zeros.

In another embodiment the actuation of the displacement means has in operation a repetitive character with a period of repetition, and the low-pass filtered, frequency dependent filtering of the error signal by the feedforward filter is time delayed for a delay period T substantially equal to the period of repetition.

Thus, any recurring disturbances to the control of the displacement means are thereby accounted for by the feedforward component. As neither the feedback nor the feedforward filter is able to foresee future disturbances, the delay period of the

feedforward actuation component enables a better and faster correction of recurring disturbances.

In a further embodiment the memory is in operation devised for storing a signal comprising a low-pass filtered signal, composed of the frequency dependent filtering of the error signal by the feedforward filter added to the output signal of the memory, wherein the output of the memory is the stored signal delayed by one delay period T. A synthesised feedforward component is thus applied with a delay of one period, thereby correcting for any recurring disturbances. The feedforward component is updated based on current observations for a better correction during the next period of repetition.

In another embodiment, the imaging system further comprises a sensor for measuring the position of the media, and wherein the error signal is based on the measured position of the media. Measuring the position of the media directly, results in a controlled system that uses the actual required quantity, being the position of the media relative to the imaging station, to base the actuation commands on. Any indirect measurements may result in a less accurate control of the required quantity. To measure the position of the media for instance an optical sensor, such as a CCD-sensor may be used, for determining the position of a media relative to a predetermined marker location.

In another embodiment the media displacement means comprises a drivable transport pinch, further comprising a sensor for measuring the orientation or the amount of rotation of the drivable transport pinch, and wherein the error signal is based on the measured position of the drivable transport pinch.

The measurement of the rotational position drivable transport pinch is less complex than a measurement of the actual position of the media, while the difference between the rotational position of the drivable pinch and the associated position of the media relative to the imaging station is relatively small if the properties of the pinch are relatively well known.

In another embodiment the media displacement means comprises a drive motor, further comprising a sensor for measuring the position of the drive motor, in particular of the drive shaft of the motor, and wherein the error signal is based on the measured position of the drive motor.

It is relatively easy to obtain the rotational position of the drive shaft of a motor. A rotational encoder disk may be fixed to the drive shaft, or an internal position encoder may be integral part of the electric motor.

In another embodiment the feedback filter comprises a proportional component acting on the magnitude of the error signal and a derivative component acting on the rate of change of the error signal.

The resulting feedback filter will result in a fast correction of incidental disturbances, while the derivative component introduces enough damping to the controlled system to overcome problems due to overshoot. In imaging systems it is undesired to oscillate a media during positioning thereof and the media should be in the correct position within a relatively small amount of time.

In another embodiment the frequency dependent filtering of the error signal by the feedforward filter is amplified with a robustness factor.

To cope with a certain degree of model uncertainties the filtered error signal, which is outputted by the feedforward filter 103 is filtered by a robustness filter 104. This robustness filter is an amplifier with an amplifying factor equal to the robustness factor. Preferably the robustness factor is a value between 0 and 1. Good results have been observed with a robustness factor of approximately 0.5, which results in a 6 dB error margin.

In another embodiment the low-pass filter imposes a phase shift when filtering. Nonzero phase low-pass filters demand less computational capacity than zero phase low- pass filters.

In another embodiment the actuation command is further composed from a parametric feedforward component based on a reference signal, comprising information about the desired position of the media. An additional parametric feedforward component decreases the time to decrease the settling time.

The parametric feedforward component may comprise a compensation for the Coulomb and / or viscous friction of the media displacement means. It may also comprise a compensation for the acceleration inertia of the media displacement means. The parametric feedforward component enables a performance improvement by incorporating system knowledge of the system that is to be controlled. The parameters

of the parametric feedforward component may be tuned in advance, e.g. after manufacturing, or alternatively during a short calibration procedure during the start-up of the apparatus.

In an embodiment the imaging station comprises a printing station for applying marking material onto the media. This may for example be based on electrographic, inkjet or laser printing principles, using for example water-based inkjet, solvent or hotmelt ink, binary toner or the like. To increase the image quality of such systems it is of high importance to position the media within very strict specifications, such as media position relative to the imaging station.

In another embodiment the imaging station comprises a scanner station for digitising image data from the media. To enable an efficient scanning process with good image quality it is very important to have a well-defined media positioning.

A preferred embodiment example will now be described in conjunction with the drawings, in which:

Fig. 1 is a schematic perspective view of a printer according to the invention;

Fig. 2A is a schematic view of a control process within the controller assembly according to the invention;

Fig. 2B is a schematic view of an alternative embodiment of the control process within the controller assembly according to the invention;

Fig. 3 is a schematic overview of the control process results.

As is shown in Fig. 1 , a rotary unit 10 of an imaging system such as a printer, e. g. an inkjet printer, comprises a feed roller 12 and a worm wheel 14 mounted for joint rotation on a common axle 16. When the rotary unit 10 is rotated in the direction of an arrow A, a sheet of a print media 18, e. g. paper, is advanced in a direction B relative to a printhead 20 along a media transport path 22. The direction B is the media transport direction or sub-scanning direction of the printer, whereas the main scanning direction C, is the direction in which the printhead 20 moves back and forth across the media transport path 22.

A worm 24 is mounted to mesh with the worm wheel 14 and is driven by an electric motor 26. A disk-type encoder 28 is mounted on a drive shaft 30 of the motor 26 so as

to detect angular increments by which the worm 24 is rotated in a direction φ. By way of example, the encoder 28 may have 500 slots, so that, utilising quadrature encoding, it is possible to detect the angular increments with a resolution of 2000 per revolution of the worm 24.

The worm gear formed by the worm 24 and the worm wheel 14 provides a very small transmission ratio 1/k « 1 , so that a relatively large angular displacement of the worm 24 leads only to a relatively small advance of the media 18. Thus, in principle, the encoder 24 permits to fine-control the media advance with very high accuracy. The number k is preferably an integer and indicates the number of turns that the worm 24 has to make for causing the rotary unit 10 to make one complete turn. Thus, when the worm 24 is rotated by 360° (a full turn), the media 18 will be advanced by a unit length δS = πd/k, with d being the diameter of the feed roller 12.

A controller assembly 50 is adapted to receive measurements from encoder 28 by means of an input module 53 and sends actuation signals to the motor 26 by means of an output module 52. A processor module 51 controls the input module 53 and output module 52. The output module 52 comprises a motor driver 52 which transforms the digital signal of the processor module 51 into a signal, such as a certain voltage, current or pulse frequency, that the motor can interpret or use directly to rotate its rotary axle 30 so as to advance the media 18 by a required length, each time the printhead 20 has performed a pass across the media 18.

The controller assembly 50 communicates with a printer controller (not shown) to determine the moment and amount of required movement of the feed roller 12. Depending on this communication a desired position or motion of the worm 24 is determined by the processor module 51.

It will be clear that alternative drive arrangements may profit from the same type of controller assembly as well. For example a direct drive feed roller, which is driven directly on the axle of rotation, or a belt driven feed roller.

Fig. 2A shows a schematic view of a control process within the controller assembly 50.

The controller assembly 50 receives a signal from the printer controller indicating the required position of the drive shaft 30. It will be clear that the printer controller may also

indicate a required position of the print media 18, of the feed roller 12, of the worm wheel 14 or any other indication of a position of a direct or indirect controlled part of the system. This indication of the required position of the drive shaft 30 is inputted in the control process as the reference signal r.

The input module 53 of the controller assembly 50 receives measurements from the encoder 28 on the drive shaft 30. This indication of the position of the drive shaft 30 is fed into the control process as the output signal y. In an alternative embodiment the position of the media 18 relative to the imaging station 20 is measured as an output. The measurements of the position of the encoder 28 are received, digitised and transformed for use in the control system in receiving unit 107. The difference between the reference signal r and the output signal y is called the error signal e. The error signal is an indication of the difference between the required position of the drive shaft 30 and the actual or measured position of the drive shaft 30.

The controller assembly comprises a feedback filter 101. This feedback filter 101 uses the error signal e to synthesise a feedback component of the actuation command u, that the output module 52 can use to drive the electric motor 26. The digital signal output module 102 sends a digital signal comprising information about the actuation command u to the output module 52 of the controller assembly. The output module 52 transforms the digital signal into a signal that the electric motor can interpret or use directly to drive the drive shaft 30.

The feedback filter 101 is a linear feedback filter and is devised to react on several properties of the error signal e. The feedback filter 101 comprises a proportional part which responds to the magnitude of the error signal e; the larger the error signal is, the larger the contribution to the actuation command will be. Thus, a large difference between the required position and the actual or measured position of the drive shaft 30 will result in a proportionally large actuation of the electric motor until the difference is smaller. The feedback filter 101 further comprises a derivative part, which responds to the rate of change of the error signal e; The larger the rate of change of the error signal e, the larger the contribution to the actuation command will be. Thus, the electric motor will be actuated more intense if the difference between the required position and the actual or measured position of the drive shaft 30 changes fast and the actuation will be smaller if the change of the error is smaller.

Alternatively the feedback filter may also comprise an integrating part, which responds to the time-integrated amount of difference between the required and the actual position of the drive shaft 30.

The process of determining an actuation command to send to the electric motor by responding to the error signal, which comprises information about the difference between a required position and an actual position may be considered as a closed-loop. This closed control loop operates at a predetermined frequency f. Depending on the operating frequency f, after each time period Ts, being equal to the inverse of the operating frequency 1/f, a new actuation command is synthesised by the feedback filter 101. The time period Ts is called the sample time of the control system. It is preferred that at least once in every sample time a new measurement of the position of the drive shaft is available.

The closed-loop-controlled drive shaft 30 has a certain closed-loop-controlled characteristics depending on the tuning of the feedback filter 101 and on the system characteristics of the drive shaft 30 itself. These characteristics determine how the controlled drive shaft 30 will react on a certain reference or sequence of references. Ideally the output of the controlled system should be instantaneously and exactly equal to the required output. In this case, the position of the drive shaft should ideally be exactly equal to the required position after each and every sample time Ts. In practice this will generally not be the case. The system needs some time to overcome the distance and this will take some time. Besides these physical limitations, in practice there may be incidental or structural irregularities, which introduce a disturbance to the output. For example, the unroundness of the drive axle, or irregularities in the worm gear may result in disturbances to the position control of the drive shaft 30.

The control assembly 50 further comprises a feedforward filter 103. The feedforward filter 103 is devised such that the closed-loop controlled characteristics of the closed- loop controlled system are compensated. The closed-loop controlled system's characteristics may be modelled by the process sensitivity Sp. This process sensitivity Sp is a transfer function that describes the relation between a certain reference or sequence of references and the output of the closed-loop controlled system.

The feedforward filter 103 is devised to equal or at least approximate the inverse of the process sensitivity Sp. Ideally the relation between the reference signal and the output of the controlled system is a one-to-one relationship, i.e. the output of the controlled

system would than be instantaneously and exactly equal to the reference. In general, the process sensitivity is not equal to one for all reference signals. By adding an additional feedforward component to the feedback component of the actuation command, which feedforward component is based on the inverse of the process sensitivity, the transfer function of the resulting feedback and feedforward controlled system is a better approximation of the desired one-to-one relationship. Feedforward filter 103 is implemented as a digital filter that equals the inverse of the process sensitivity Sp of the controlled system. The process sensitivity Sp of the controlled system or an approximation thereof may be measured directly, but may alternatively also be constructed theoretically, by modelling or measuring the transfer functions of the feedback filter and the system or process that is to be controlled. The process sensitivity that is used for designing the feedforward filter 103 is constructed from a theoretical modelling of the controller and frequency response measurements of the electrically driven feed roller 12. To cope with a certain degree of model uncertainties the filtered error signal, which is outputted by the feedforward filter 103 is filtered by a robustness filter 104. This robustness filter is an amplifier with an amplifying factor between 0 and 1. To incorporate robustness against 6 dB model-uncertainties the robustness filter 104 is set to 0,5. The modelling and frequency response measurements of the process sensitivity of the electrically driven feed roller 12 are accurate for lower frequencies but become increasingly less accurate for high frequency effects. Nevertheless, inverting the process sensitivity Sp for use in the feedforward filter 103 increases the influence of the high frequency effects, which are determined with a relatively low degree of accuracy. Therefore, the filtered error signal that is outputted by the feedforward filter 103 is fed through a low-pass filter 105, which filters out all signals above a predetermined frequency. This frequency is called the cut-off frequency. Because high frequency actuation of the drivable feed roller 12 does not have a significant influence on the controlled system, and because the high frequency modelling of the feedforward filter is less accurate, the low pass filtering of the feedforward component of the actuation command does not deteriorate the controlled system.

The low-pass filter is implemented as a zero phase low pass filter, thus the low-pass filter imposes no phase shift on the signal when filtering. The reference signal of the imaging system, in particular the reference signal of the displacement means, e.g. the feed roller has a highly repetitive character. After each

scanning movement of the printhead 20 in main scanning direction C the media is advanced in transport direction B. To advance the media accurately over a predetermined distance δS, the worm 24 is rotated over exactly one complete revolution, i.e. 360°. Driving the worm 24 for a full revolution after each swath of the printhead 20 is a highly repetitive reference signal with a period of repetition Tr.

Neither the feedforward filter 103, nor the feedback filter can foresee future events. Disturbances that occur during each repetition of the controlled movement, such as unroundness of the drive shaft 30 or irregularities of the worm 24 or worm wheel 16 can only be acted upon after they have occurred and after they have been detected by the position sensor 28.

A memory 106 is implemented, which is devised to store a signal comprising the low- pass filtered signal, composed of the frequency dependent filtering of the error signal by the feedforward filter 103 added to the output signal of the memory 106 itself, wherein the output of the memory 106 is the stored signal delayed by one delay period, equal to the period of repetition Tr. An actuation command that was calculated to correct for an error in the previous repetition will therefore be applied during the next repetition of the controlled drive shaft motion. The feedforward filter 103 therefore accounts for repetitive errors, while the feedback filter 101 accounts for incidental errors.

Fig. 2B shows a schematic view of an alternative embodiment of a control process within the controller assembly 50. The low-pass filter 1 15 is implemented as a non-zero phase low-pass filter. Such low-pass filter 1 15 does impose a phase shift on the signal, but requires less computing capacity with respect to the zero phase low-pass filters. A phase shift on the control signal may slightly deteriorate the actuation command, but an additional parametric feedforward filter 110 compensates the slight deterioration. The parametric feedforward filter 110 acts on the reference signal r and contributes an additional component to the actuation command. This component comprises a compensation for the Coulomb and viscous friction of the controlled system and compensates for the acceleration inertia of the media displacement means. As these system properties of the controlled system are not expected to change significantly during operation, these compensations can be tuned in advance, or during a short calibration procedure at the start-up of the imaging system. The combination of the parametric feedforward filter 1 10 and a non-zero phase low-pass filter 115 result in smaller computational demands to the processing module 51.

Fig. 3 shows a schematic overview of the control process results in repetition one (I), two (II), three (III) and ten (X). As shown in the first row, the reference in this example is a sine-shaped signal. The controlled system is required to follow a sine-shaped signal formed reference signal. In the second row the periodic disturbance has been illustrated. This block signal disturbance is imposed in addition to the actuation command. This means that the controlled system applies a combination of a calculated actuation command and the block signal disturbance. The physical reason for this disturbance is irrelevant for this example. In the sixth row the measured output of the system has been depicted (solid line) and the reference signal (dashed) has been added for illustrative reasons. The influence of the block disturbance is clearly visible in the first period (I). The error signal formed by the difference between the reference r and the output y is depicted in row three. This error signal is clearly influenced by the disturbance and furthermore comprises sine- shaped influences of the inherent time lag caused by e.g. the inertia of the rotating parts such as the feed roller 12.

With reference to the first period (I) it is clear that the feedback component (shown in row four, u _fb ) acts on the actual error signal, while the feedback component (shown in row five, u _ff ) has no effect yet. With reference to the second and third period, as shown in columns two (II) and three (III) it is noted that the feedforward component of row five (u _ff ) now clearly incorporates a part of the sine-shaped feedback command of the first period and further an inverse block-shaped part has been synthesised to compensate for the block-shaped disturbance as detected in the previous period. This trend in increased in the third period and results in a decreasing overall error. Hence, the feedback component is decreased while the tracking performance of the system, i.e. the capability to follow the reference is maintained, or even improved.

After ten periods of repetition (period X) it is clear that the tracking performance is very good, the error approaches zero, the feedforward component has been synthesised to correct for the block-shaped disturbance and the repetitive actuation of the system, while the feedback component corrects for incidental errors only.