GB1563032A - Method of checking strips of material - Google Patents

Description

(54) A METHOD OF CHECKING STRIPS OF MATERIAL (71) We, G.A.O. GESELLSCHAFT FUR AUTOMATION UND ORGANISA TION m.b.H., a German limited liability company, of Euckenstrasse 12, Munich, Federal Republic of Germany, do hereby declare the invention, for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement: The present invention relates to a method of checking a strip of material provided with periodically recurring patterns for faults, wherein the strip area to be checked is scanned in parallel tracks.

It is known from German Patent No.

2,322,803 to check for faults in a paper web consisting of two or more tracks concurrent with respect to the sequence of areas of different transparency. In that method, with which particularly bank-notes provided with watermarks are checked, the web is scanned along two tracks running parallel to the direction of transport with a row of photocells extending transversely to the direction of running of the web, the output signals of each pair of photocells associated with corresponding places in the two tracks being compared by means of a differential circuit.

In the case of faultless webs. since the corresponding places of the concurrent sequences of marks should have the same transparency, a difference signal is obtained which is within a preset threshold. When the difference signal exceeds the preset threshold limits, this is a sign that one of the two compared tracks is fault.

A disadvantage of this method is that it is impossible to check a periodic pattern which is present in the web only once, because two parallel transversely spaced tracks are required for comparison one with the other.

It is an object of this invention to enable strips of material having periodically recurring patterns to be checked in such a manner as to meet various quality requirements in an economic manner and to permit a single track to be checked.

According to this invention there is provided a method of checking a strip of material having a periodically recurring pattern, the method comprising the steps of: scanning the strip area to be checked in a plurality of parallel tracks; dividing each indicidual track into a plurality of sections by periodically sensing a corresponding scanning output signal in accordance with a clock signal synchronised with the recurring pattern; allotting threshold values to the sections, which values are determined in response to the scanning output signals; storing the threshold values; comparing subsequent scanning output signals with the stored threshold values; and indicating a fault when a threshold value is exceeded.

In a preferred embodiment of the invention, the threshold values of corresponding sections of a preset number of patterns is averaged. To compensate for permissible long-term variations which are to be ignored, the threshold values obtained by averaging are updated with newly determined values at certain intervals.

This method proves to be particularly advantageous in that, with given track width and number of tracks, a wide range of checking resolution can be obtained by variation of the length of the sections.

If, for example, only faults occurring outside the pattern and faults in pattern spacing are to be determined (which may be the case especially with patterns occupying a small percentage of the web area), a threshold value slightly exceeding the noise level due to, e.g., the cloudiness of the paper, will be generated for the vellum portion, i.e.. the pattern-free areas of the web. In contrast, the threshold values for each individual track in the pattern area are raised along the whole length of the pattern to such an extent that the largest amplitude value of a pattern found to be correct will still be below the threshold values.In this manner, only those faults in the patterns which stand out from the general printing and impression level can be detected, the faults in the pattern-free area, which are much more noticeable in most cases, being detected relatively reliably.

If, however, a checking accuracy is desired which permits even the small deviations in the pattern to be detected, the pattern area is divided into as many sections as possible so that, in the ideal case, each section will be allotted at least one threshold value. In this manner, an envelope is placed over the output signals of the pattern areas which is more or less exact according to the length of the sections and with which the congruence of the periodically recurring track areas can be monitored. In order that the "good" output signals lying partly above the arithmetic mean are not regarded as fault signals, the threshold values are raised above the output signal by adding a constant voltage value. Here too however, the vellum portion is checked with a uniform threshold.

Since the faults in the amplitude range of the output signal may assume any value, in the positive amplitude range for example, an otherwise positive section value may assume not only a smaller, but even a more negative value as a result of a fault. Such "oppositely directed" faults are detected by opposite polarity threshold values.

To limit the effect of oppositely directed faults in the determination of the averages serving for a comparison, it is advisable to admit only the amplitude values of less than zero potential.

The number of sections in the area of the patterns can be fixed according to different aspects. If. for example, pattern areas having a small positional and pattern tolerance.

such as print patterns, are to be checked, the checking accuracy can be considerably increased by making the number of sections as large as possible so that even the smallest deviations in the pattern can be detected. In this case, the setting is optimal if the section duration of the half-wave length corresponds to the smallest detectable fault, i.e.. if the setting is dependent on the highest frequency.

To check watermarks, a different setting is to be recommended. Watermarks have greater manufacturing tolerances than those of print patterns, for example. With a greater positional tolerance. short-time signal amplitudes of corresponding sections would not give a continuous envelope.

Taking into account the permissable positional deviations, the sections must therefore be chosen so large that, at the maximum positional deviation, corresponding points of successive watermarks can still be covered by one section. The greater the permissible positional deviation, the longer the length of the sections, and the coarser the envelope.

With a transmitted-light check, the output signal required for evaluation can be made available either in the form provided by photocells, which also contains the average thickness of the paper as a d.c. voltage component, or as a differentiated photocell signal. Since the average paper thickness is mostly checked already during the paper manufacture, and since the differentiated signal characterizes changes of contrast by considerably faster signal changes, and since such a signal automatically has a zero d.c.

level, preference will be given to the differentiated signal.

A further improvement in fault detection is achieved by a special signal processing technique. The frequency spectrum of a translucent web without patterns can be divided into three basic frequency ranges.

The first range (I) contains predominantly the long-wave thickness changes of the paper, and the third range (III) contains the high-frequency amplitudes resulting from sharply defined brightness differences, e.g.

from small faults. The middle range (II) contains predominantly the background noise or cloudiness of the paper. In the second frequency range. therefore, faults can be detected only if their amplitude exceeds the general noise level. By suitable filtering. faults can thus be separated according to the following criteria: a) abruptness of the fault edges (III) b) amplitude of the faults (II) c) thickness deviations (I).

According to the application, the number of sections and the frequency range for the evaluation of the output signals can thus be chosen freelv. Also, various combinations of the different web areas are possible. If, for example. general thickness changes of paper knots or the like are to be detected, the check will be carried out in the first frequency range (I) with only one section in the unmarked area. In contrast, water-mark areas will be checked with the aid of the second and third frequency ranges taking into account the positional tolerance for the section length.

The invention will now be described by way of example with reference to the accompanying drawings. in which: Figure 1 is a schematic representation of a transmitted-light scanning device operable in accordance with the invention and appropriate for watermarked paper; Figure 2 shows a web with an indicated pattern area; Figure 3 shows an output signal from a scanned track with a raised threshold value; Figure 4 shows an output signal from a scanned track, with a plurality of threshold values in the region of a patterned area; Figure 5 is a block diagram of a scanning device operable in accordance with the invention; and Figure 6 is a simplified circuit diagram of an averaging circuit.

Figure 1 shows the basic construction of a checking device in accordance with the invention which is suitable for web checking. Its operation will be explained with the checking of a bank-note web provided with watermarks. This checking station is located in front of the cross cutter and sorter which cuts the web into individual sheets at a correct distance from the marked area and sorts out faulty marked sheets by means of a sorting gate. It is obvious to the person skilled in the art that this is just one of many examples of application, and that print patterns and the like can be checked in the same manner.

A web 1 being moved over feed rollers 2, 3 in the direction of an arrow 4 has watermarks which recur periodically in the direction of transport and whose location and structure are to be checked, together with the areas free from watermarks (vellum portion), for deviations and faults. Below the web 1, there is provided a rod-shaped lamp 5 as a light transmitter which homogeneously transilluminates the paper over its entire width. On the opposite side of the web, there is provided a row of photocells 6 as receptors. If the web 1 moves continuously, the light passing therethrough varies in brightness because of the different transparency. The individual photocells convert these variations in brightness into amplitude variations of their electrical output signals.The individual analog output signals of the photocells are applied to a signal conditioning circuit 7 where thev are amplified and changed into a form necessary for further processing. Also provided are a clock generator 8, a cross mark sensor 10.

and a mark generator 9.

Since the lamp 5 and the row of photocells 6 are arranged transversely to the direction of transport, the web 1 being moved continuously in the direction of the arrow 4 and, consequently, the watermarks 11 impressed on the paper, if e photocells are present, is divided into a, - a, parallel tracks 12 extending in the direction of transport. as shown in Figure 2. Since these watermark areas 14 on the web 1 generally recur at regular intervals. the same output signals of the photocells will recur in a periodic sequence, exceeding the noise level of the vellum portions in amplitude by a multiple.

To be able to check these recurring watermark signals of the individual tracks 12 by the method according to the invention, the a1 - ac individual tracks 12 are subdivided into bl - bn sections 13. Thus, an imaginary two-dimensional checking grid is superimposed on the watermark areas 14. Its fineness is dependent on the length of the sections 13 and on the number of tracks 12 (Figure 2).

This requires a paper-transportdependent clock signal from the clock generator 8 and a watermark-positiondependent sync pulse from the cross-mark sensor 10. In practice, to limit the sheets, the web 1 is provided with a so-called cross mark (watermark) for each sheet; this mark, detected by a photocell, is also used to synchronize the sections 13. The dimensions of the watermarks, i.e., length and width, are manually entered into a programming unit.

By separate determination of the highest and lowest amplitude values in the output signals of the photocells for the individual tracks 12, and by the assignment of threshold values thereto, each track is allotted an upper and a lower envelope, i.e., one for the positive amplitude range of the output signal and one for the negative amplitude range, for the comparison with subsequent patterns. The envelopes thus represent thresholds above or below which the output signal must not pass at any point if the web material is free from faults.

Figures 3 and 4 show two output signals 15 obtained by scanning one of the tracks 12, along with the associated envelopes. The relatively greatly varying opacity conditions in the watermark 11 are represented in a range 17 of these output signals 15, whereas the more uniform output signal sections of the vellum portion, which are caused by the cloudiness of the paper, are allotted to a range 16.

Figure 3 shows a division of the tracks for less stringent quality requirements with which primarily only the particularly disturbing faults in the vellum portion are to be detected and indicated. The sections are chosen to be so large that the output signal 15 is limited only by a single upper positive and a single lower negative threshold value during a watermark in the range 17, and by a single positive and a single negative threshold value during the vellum portion in the range 16. The threshold values obtained by determining the maximum amplitudes are indicated by lines 22. The hatched areas 20 mark the voltage value added as a tolerance range. By means of the thresholds 21 so formed and relatively closely adjacent in the vellum portion, opacity changes in the area free from pattern are detected with high accuracy.Faults in the watermark which generate amplitudes smaller than the respective largest amplitude are not detected because of this very coarse grid. Since minor brightness differences in the watermark are far less disturbing than those in the vellum portion, the quality of a web 1 so checked is sufficient for many cases despite the relatively small amount of checking apparatus required.

For stringent requirements, and particularly to permit subsequent mechanical verifications of bank notes, however, it is recommended in the case of bank-note paper to change the thresholds 21 above and below the watermark 11 so that even minor brightness differences in the watermark can be detected as faults and singled out. This is done in the pattern area by "clocking" the tracks into considerably smaller sections. In Figure 4, for example, the watermark area is divided into 15 like sections bl - b15. Like in the simplified method shown in Figure 3, the highest and lowest amplitudes are determined in each section, and the threshold values are generated and are each allotted one threshold 21 increased or decreased by a fixed voltage value. In spite of the still rather coarse grid, unlike in the example of Figure 3, a relatively good match of the output signal can be recognized.

It is easy to see that the comparison of output signals with such envelopes shows far more faults, and that the checking accuracy can be increased with increasingly fine clocking up to the exact proof of congruence.

An optimization of the section length can be achieved particularly in the case of very small faults if the length of the sections corresponds to the half-wave length of the smallest fault to be detected. To obtain safely a continuous envelope during averaging, however, the section length should not be smaller than the maximum permissible positional variation and/or distortion deviation. In the case of very small tolerances, the length of the sections will therefore be adapted to the half-wave length of the smallest fault to be detected. If greater tolerances have to be taken into account.

the length of the sections will be selected on the basis of the maximum permissible variation.

Figure 5 shows a block diagram of a one-track checking device working by the method according to the invention: the arrangements for the other tracks are indicated.

The checking device comprises essentially (i) a programming unit 25 in which the actual watermark position or register" is stored and which is controlled by the clock generator 8 and by the cross-mark sensor 10; (ii) a preamplifier 24 which amplifies the output signal of a photocell 23 of the row 6 of photocells and, if necessary, divides it into the desired frequency ranges by means of suitable filter circuits; (iii) a storage and arithmetic unit 26 which generates and stores the threshold values; (iv) a control unit 27 which controls the time sequence of data acquisition, storage, and reproduction; and (v) a comparing unit 28 which compares the respective valid threshold values with the output signal of the photocell 23 and indicates the transgression of a threshold as a fault.The above-mentioned preamplifier 24 and the units 25 to 28 as well as an OR gate 69 to be explained below are contained in the signal processing circuit 7 (Figure 1).

The programing unit 25 and the control unit 27 need be provided only once per checking device, since they control the entire installation, whereas the storage and arithmetic unit 26, the comparing unit 28, and the preamplifier 24 must be present once for each track to be checked. With 4 tracks 12, the checking device thus consists of 4 photocells 23 in the row of photocells 6, 4 comparing units 28, 4 storage and arithmetic units 26, one programing unit 25, one control unit 27, the clock generator 8, and the cross-mark sensor 10. If a web is to be checked according to several separate criteria, one storage and arithmetic unit 26 and one comparing unit 28 must be provided for each track 12 and for each frequency range, as is indicated in Figure 5 by the blocks 26a, 26b, 28a. and 28b.

The web 1 to be checked, on which, as mentioned above, the individual watermarks 11 are lined up to form so-called registers, and on which cross marks are provided for synchronization, passes through the checking device in the manner shown in Figure 1 and is scanned for faults by the row 6 of 1 photocells 23. With a photocell 32 and a lamp 33 in the cross-mark sensor 10, the sync marks in the paper are determined. The photocells 23 provide the output signals corresponding to the output signals 15 shown in Figures 3 and 4, while the photocell 32 delivers a synchronizing pulse only at the beginning of a register. In addition, the clock generator 8. driven by the transport system and consisting of a photocell 29, a slotted disk 30, and a lamp 31, provides a square-wave signal which is permanently associated with the pass of the paper and is used to "clock" the tracks 12.

According to the desired length of the sections 13, several clock periods are combined into sections of the length b in a circuit 34 included in the programing unit 25. The term (N3) in the graphical symbol indicates that each section 13 of the length b is obtained by combination of 3 clock periods. For each type of paper, the length of the sections and the time the formation of the sections in the registers sets in can be manually entered into a matrix circuit 35 as a checking program. The matrix circuit 35 is controlled by the cross-mark sensor 10 and, in turn, controls the circuit 34.

Over a clock line 36, the clock signal adapted to the section lengths is sent from the circuit 34 to a store pulse generator 38 and a store address generator 39. From the store pulse generator 38, the signals formed in a rectifier circuit 46 are then chopped up into individual sections. The rectifier circuit 46 includes a peak-value meter-type rectifier of known design. This rectifier circuit 46 may also contain a differentiating circuit.

Preferably, the differentiated signals are processed. In addition, the store pulse generator 38 impresses on a summer 49, analog stores 50, 52, and a comparator 55, over a line, the clock rhythm necessary for averaging and for the comparison. With each section pulse, the generator 38 generates a pulse sequence which, as will be described in more detail below, controls essentially the read and write cycles of the stores 50 and 52. In the meantime, the store address generator 39 is advanced by one store address with each section pulse for the control of the stores 50, 52. as will also be explained in more detail below. The term (l) in the generator 39 means for practical application that n store addresses are present which are successively coded out one at a time.

A clock line 37 coming from the matrix circuit 35 supplies a pattern counter 41 in a similar manner with clock pulses which mark the beginning and the end the individual patterns. In the pattern counter 41, the number of pattern areas used for averaging is preset. It therefore provides the signals which, after a preset number n of patterns has been attained, cause the old threshold values stored in the store 52 to be exchanged for the new ones formed in the store 50. The pattern counter 41 thus counts the successively scanned watermarks and. if in agreement with the presetting, provides a coincidence signal which transfers the store contents from the store 50 to the store 52 via an AND gate 51 and resets the generators 38, 39 via a control line 42.The control unit 27 also includes a logic circuit 43 which registers any errors during the averaging via a line 45 and influences the pattern counter 41, thus preventing the overwriting of the new envelope as a function of the number or magnitude of the errors. because otherwise false threshold values would be used to check the subsequent watermark series.

For the sake of simplicity. the storage and arithmetic unit 26 will be explained with reference to the formation of a positive envelope by means of the rectifier circuit 46, the summer 49, and the store 50. The negative envelope is formed in a similar manner with like circuit elements, indicated in the lower half of the unit 26.

The output signal of the photocells 23 passes through the preamplifier 24 to the rectifier circuit 46, where the largest amplitudes within the sections are determined and where threshold values dependent thereof and, consequently, thresholds 21 are formed, and to a buffer amplifier 54, behind which it is compared with the threshold value contained in the store 52. The output signal of the rectifier circuit 46, which has for each section a voltage value dependent on the respective highest amplitude, is applied to the summer 49 via a line 47 and a quotient circuit 40. Controlled by a preselector switch 44 and as will be described in more detail below, the quotient circuit 40 divides each voltage level allotted to the sections down to the mth value, where m corresponds to the preset number of patterns.This voltage value is added in the summer 49 to the previously determined values of the same sections for averaging, and is stored in one out of n cells of the store 50 assigned to the respective section.

Accordingly, after m watermarks have been checked, the m cells of the store 50 contain the averages of the individual sections found of m watermarks. At the end of mth check, the pattern counter 41 applies to the input 53 of the AND gate 51 a voltage pulse which enables this AND gate 51. Thus, after the previous contents have been erased, the contents of the store 50 are successively transferred to the store 52 over a line 67. To ensure that, during the individual storage processes. the individual voltage levels are always entered into the proper cells of the stores 50 and 52, and that during the comparison of the actual levels, the corresponding threshold values of the store 52 are always available, the store address generator 59 successively opens and closes the individual cells of the stores 50 and 52 in the rhythm of the signals provided by the circuit 34.

The voltage levels stored in the store 52 are used for the continuous comparison with the instantaneous levels provided by the buffer amplifier 54. They are applied to inputs 57 and 58, respectively, of the comparator 55, which provides at its output 59 a fault signal when the voltage value applied to the second comparator 58 exceeds that applied to the first input 57. This fault signal is passed to an output terminal 70 through OR gates 68. 69. By means of the fault signals at the output terminal 70, the mark generator 9 indicated in Figure 1 can then be operated. A comparator 56 performs the same function and evaluates the signal for the negative amplitude range and also sends its fault signal to the output terminal 70 via the OR gate 68, 69.

Since no unambiguous comparison between the signals provided by the buffer amplifier 54 and the values stored in the store 52 is possible during the transfer of the store contents from the store 50 to the store 52, the comparators 55 and 56 are inhibited for these periods by the generator 38 via the line 60. The delivery of a fault signal based on undefined voltage conditions during the transfer is therefore impossible. Since the transfer takes only fractions of the actual clock time, the resulting blankings are negligible.

The averaging performed in the storage and arithmetic unit 26 will now be explained again in greater detail with the aid of the schematic diagram shown in Figure 6.

When the section bl of the first of m watermarks 11 (Figure 2) is checked, the voltage value conditioned by the rectifier circuit 46 is applied to the quotient circuit 40 over the lead 47. With the aid of the preselector switch 44, which is also mechanically coupled with the pattern counter 41, the division ratio of the quotient circuit 40 is set so that only the corresponding portion of the applied voltage becomes effective at the input of an operational amplifier 61 included in the adder 49. The resistors 62 and 63 have a ratio of 1:1, so the full voltage value of an intermediate storage cell 64 becomes effective at the adder input. Since a switch 65 formed by a field-effect transistor (FET) first remains off during the first watermark, a i0' potential appears at the adder input.Thus, the mth part of the voltage value a1 appears at the output of the operational amplifier 61. By opening a switch 66 formed bv a field-effect transistor (FET), this voltage value is written into the cell S1 of the store 50, which cell was opened before bv the store address generator 39. In a similar manner, the voltage values of the sections b2 - bll are transferred to cells S1- S,.

The threshold value formed for the vellum portion is stored in the cell Son+,. This threshold value, like the individual threshold values of the envelope. is determined by ascertaining the highest amplitude. Since. as mentioned above. the check of the entire vellum portion is performed with only one fixed threshold, in the matrix circuit 35 the entire vellum portion between two watermarks is equated with one section by suitable programing.

At the second of the m watermarks. at the beginning of each section. the intermediate storage cell 64 is successively loaded. via the switch 65, with the voltage value present in the respective cell S of the store 50. and this value is added to the voltage levels provided by the quotient circuit 40. At the output of the operational amplifier 61, the newly arriving voltage levels are thus added to the associated stored values of the same sections and entered again into the cells S via the switch 66. After passage of m watermarks, the cells Sl ...Sn assigned to the sections have therefore been loaded with the averages taken over m watermarks. These averages are transferred to the store 52 over the line 67 prior to the next, i.e. (m+1) watermark, as explained above.During the check or averaging and during the transfer, the stores 50 and 52 are addressed with the same addresses, i.e., the same cells are opened and closed in synchronism. While the comparing unit 28 is now comparing the next m watermarks with the levels of the store 52, the store 50 is being loaded with new averages. In this manner, the checking device is provided relatively economically with new threshold values during the check without interruption of the checking process except for the negligible blanking during overwriting.

As mentioned at the beginning, the individual averages are raised by a fixed amount. This is done by means of a simple summing circuit consisting of an operational amplifier and a few resistors. For the sake of simplicity. this unit is not shown. It is included in the store 52, i.e., connected ahead of the store output.

When checking the surface of patterns with a uniform threshold (see Figure 3), the amount of circuitry required can be considerably reduced. Instead of being passed through summing and memory circuits, the signal coming from the rectifier circuit 46 is passed through a less expensive low-pass filter and compared with the output signal of the buffer amplifier 54. With this modification of the invention. a fault causes an immediate change in the threshold used for the comparison. With a time constant corresponding to the passage of, e.g., five watermarks, however, this single fault causes only a short-time change in the threshold value by about s of the actual value. Fault detection is therefore affected only very little although the amount of circuitry required is greatly reduced by the elimination of the need for the summer 49, the stores 50, 52. and the AND gate 51. A further simplification can be achieved if only the positive signal components are taken into account in the check. By the rectification of the signals coming from the preamplifier 24, preferably with a full-wave or bridge rectifier, whereby the negative signal components are folded upwards, the resulting deterioration in quality can be kept within tolerable limits.

Without departing from the principles according to the invention, the web 1 may, of course, also be checked transversely to the direction of transport instead of with longitudinal tracks 12. The tracks are then no longer scanned parallel but serially, e.g.

with a keyed row of photocells or a light beam deflected by a mirror wheel or the like. Nevertheless, the fundamental idea of the section and threshold-value assignment can be fully adopted.

WHAT WE CLAIM IS: 1. A method of checking for faults a strip of material having a periodically recurring pattern, the method comprising the steps of: scanning the strip area to be checked in a plurality of parallel tracks; dividing each individual track into a plurality of sections by periodically sensing a corresponding scanning output signal in accordance with a clock signal synchronised with the recurring pattern; allotting threshold values to the sections, which values are determined in response to the scanning output signals; storing the threshold values; comparing subsequent scanning output signals with the stored threshold values; and indicating a fault when a threshold value is exceeded.

2. A method according to claim 1, wherein the parallel tracks to be scanned run lengthwise in the strip of material.

3. A method according to claim 1, wherein the strip of material is transported relative to scanning apparatus, and wherein the parallel tracks to be scanned run transversely to the direction of transport of the strip.

4. A method according to any of claims 1 to 3, wherein the length of the sections in the individual tracks corresponds to the length of a pattern.

5. A method according to any of claims 1 to 3, wherein the individual tracks. in the area of each pattern, are divided into several sections of the same length, and that separately determined threshold values are allotted to the individual sections.

6. A method according to any of claims 1 to 3, wherein the length of the sections is fixed as a function of the positional tolerance between the marked areas in the patterns.

7. A method according to any of claims 1 to 3, wherein the length of the sections is fixed as a function of an output signal pulse width corresponding to a smallest permissible fault.

8. A method according to any preceding claim, wherein the individual threshold values are formed by adding a fixed voltage value to the highest amplitudes present in the respective sections.

9. A method according to any preceding claim, wherein the formation of the threshold values in the individual sections is effected after the scanning of a predetermined number of patterns.

10. A method according to claim 9, wherein the threshold values determined after a predetermined number of patterns are used to check the subsequent same number of patterns.

11. A method according to any preceding claim, wherein the threshold values of the individual sections are replaced by new threshold values after a predetermined number of patterns has been checked.

12. A method according to claim 11, wherein the new threshold values are determined during, or parallel to, the checking of the predetermined number of patterns.

13. A method according to any of claims 9 to 13, wherein the threshold values of homologous sections of the predetermined number of patterns are averaged.

14. A method according to any of claims 1 to 8, wherein the stored threshold values of individual sections are changed immediately by the homologous output signal components of the instantaneous patterns.

15. A method according to claim 14, wherein the speed of the threshold-value change is adjustable by means of delay lines.

16. A method according to claim 1, wherein the stored threshold values are compared with the differentiated output signals of the current scanning.

17. A method according to claim 16, wherein rectification of the output signals of the current scanning is performed by means of a full-wave or bridge rectifier.

18. A method according to any preceding claim, wherein the scanning output signals are split up into at least two frequency ranges, and the checking for one or more of these frequency ranges is performed separately.

19. A method of checking for faults in a strip of material having recurring patterns, wherein the strip to be checked is scanned in parallel tracks, the scanning output signal from each track being compared with a stored value representing a corresponding scanning output signal or signals obtained during a given time from the same track on a fault-free portion of the strip, which time interval is dependent on the rate of recurrence of patterns in the strip.

20. A method of checking strips of material substantially as herein described with reference to the drawings.

**WARNING** end of DESC field may overlap start of CLMS **.

Claims (20)

**WARNING** start of CLMS field may overlap end of DESC **. of course, also be checked transversely to the direction of transport instead of with longitudinal tracks 12. The tracks are then no longer scanned parallel but serially, e.g. with a keyed row of photocells or a light beam deflected by a mirror wheel or the like. Nevertheless, the fundamental idea of the section and threshold-value assignment can be fully adopted. WHAT WE CLAIM IS:

1. A method of checking for faults a strip of material having a periodically recurring pattern, the method comprising the steps of: scanning the strip area to be checked in a plurality of parallel tracks; dividing each individual track into a plurality of sections by periodically sensing a corresponding scanning output signal in accordance with a clock signal synchronised with the recurring pattern; allotting threshold values to the sections, which values are determined in response to the scanning output signals; storing the threshold values; comparing subsequent scanning output signals with the stored threshold values; and indicating a fault when a threshold value is exceeded.

2. A method according to claim 1, wherein the parallel tracks to be scanned run lengthwise in the strip of material.

3. A method according to claim 1, wherein the strip of material is transported relative to scanning apparatus, and wherein the parallel tracks to be scanned run transversely to the direction of transport of the strip.

4. A method according to any of claims 1 to 3, wherein the length of the sections in the individual tracks corresponds to the length of a pattern.

5. A method according to any of claims 1 to 3, wherein the individual tracks. in the area of each pattern, are divided into several sections of the same length, and that separately determined threshold values are allotted to the individual sections.

6. A method according to any of claims 1 to 3, wherein the length of the sections is fixed as a function of the positional tolerance between the marked areas in the patterns.

7. A method according to any of claims 1 to 3, wherein the length of the sections is fixed as a function of an output signal pulse width corresponding to a smallest permissible fault.

8. A method according to any preceding claim, wherein the individual threshold values are formed by adding a fixed voltage value to the highest amplitudes present in the respective sections.

9. A method according to any preceding claim, wherein the formation of the threshold values in the individual sections is effected after the scanning of a predetermined number of patterns.

10. A method according to claim 9, wherein the threshold values determined after a predetermined number of patterns are used to check the subsequent same number of patterns.

11. A method according to any preceding claim, wherein the threshold values of the individual sections are replaced by new threshold values after a predetermined number of patterns has been checked.

12. A method according to claim 11, wherein the new threshold values are determined during, or parallel to, the checking of the predetermined number of patterns.

13. A method according to any of claims 9 to 13, wherein the threshold values of homologous sections of the predetermined number of patterns are averaged.

14. A method according to any of claims 1 to 8, wherein the stored threshold values of individual sections are changed immediately by the homologous output signal components of the instantaneous patterns.

15. A method according to claim 14, wherein the speed of the threshold-value change is adjustable by means of delay lines.

16. A method according to claim 1, wherein the stored threshold values are compared with the differentiated output signals of the current scanning.

17. A method according to claim 16, wherein rectification of the output signals of the current scanning is performed by means of a full-wave or bridge rectifier.

18. A method according to any preceding claim, wherein the scanning output signals are split up into at least two frequency ranges, and the checking for one or more of these frequency ranges is performed separately.

19. A method of checking for faults in a strip of material having recurring patterns, wherein the strip to be checked is scanned in parallel tracks, the scanning output signal from each track being compared with a stored value representing a corresponding scanning output signal or signals obtained during a given time from the same track on a fault-free portion of the strip, which time interval is dependent on the rate of recurrence of patterns in the strip.

20. A method of checking strips of material substantially as herein described with reference to the drawings.