WO2008122444A1

WO2008122444A1 - Method for the measurement of optical spectrums in web-fed printing

Info

Publication number: WO2008122444A1
Application number: PCT/EP2008/002844
Authority: WO
Inventors: Matthias Riepenhoff
Original assignee: Wifag Maschinenfabrik Ag
Priority date: 2007-04-10
Filing date: 2008-04-10
Publication date: 2008-10-16
Also published as: EP2134547B1; ATE485164T1; EP2134547A1; DE102007016980A1; DE502008001594D1

Abstract

The present invention relates to a method for the measurement of optical spectrums in web-fed printing, characterized in that continuous measurement is carried out of the printed web.

Description

Method for measuring optical spectra in web-fed printing

Modern printing presses are increasingly being equipped with control systems to reduce waste and ensure quality. Examples: cutting registers, color register systems, color density control. In sheetfed and roll-fed commercial printing, control elements are printed to assess print quality. In newspaper printing, it is desirable to dispense with control elements or make them unattractive.

In a method known from DE 101 31 934 A1, the measuring head of an optical measuring system is moved successively to different locations across the web. While the printed web passes the measuring head is continuously measured. The individual elements of a burst correspond to smaller image sections. The image sections are assigned to individual elements of a burst. Thus, image areas with higher area coverage can be found, by means of which a more accurate determination of the solid density can be made.

One way to implement the method uses a spectrometer and continuous illumination. The remission spectra of the moving, printed web are continuously measured.

Problems are caused by the slow readout of the spectrometer. The cells of the CCD chip in the spectral sensor are read out serially and converted into digital data. The AD conversion limits the readout speed to ts = l.365ms per part measurement ¹ . This

¹ The value applies if all cells of an example sensor are used. It results from the sampling rate of the A / D converter (187.5kHz) and the number of cells. You get: 256 / 187.5kHz = 1.365333ms. If not all cells are used, the following apply limits the scanning speed and leads to a spectral "smearing" of the data The first values (short wavelengths) correspond to the measuring spot around a pixel (X, Y), the last values (near infrared) have to be the pixel (X, Y + ts * At a web speed of 14m / s, the web travels a distance of almost 2cm during the reading time, so the measurement can not be assigned to a scene, but is "smeared" over about 2cm. With a radius of the measuring spot rm of approx. 2-4mm, this deviation is not negligible.

A simple solution is to select a spectrometer in which this problem does not occur, e.g. Frame transfer technology for CCD sensors. However, there are advantages of the serially operating sensors, so a solution for this type is desirable.

Preferred Solution: Previously, burst measurements were made so that one burst corresponded to the slice length, or the measurement time was one print cycle (or multiple). If a burst of b corresponded to partial measurements of a circumference, then for a measurement over k printing cycles, the partial measurements at time ti and tj _{+ b} correspond in each case to the same image detail. For the preferred solution it is now proposed to adjust the measuring time so that the partial measurements violate this rule. There is a beating of the measured spectra with the frequency 1 / (T- (Vt ₀ )) - 1 / T. The readout delay for the individual spectral ranges is known. One can re-sort the proportions of the spectra, so that corrected spectra arise, which can be assigned to an image area with sufficient accuracy.

FIG. 1

The section length of the machine is U. The machine speed is V. The cycle time for a printing operation is T = OfV. This corresponds to a local smearing of the spectrum by L = V * ts. This "smear length" L will be in k parts like that

Considerations a correspondingly smaller ts. For example, if only 180 cells are used, ts = 180 / 187.5kHz = 0.96ms. divided that the pitch L / k is small compared to the radius R of the measuring spot. Values of k can only be meaningful if the spectrometer has more than k support points in the spectrum. The following is about finding a value D that approximates L / k.

In a burst measurement, a single part of the burst has a measurement time tm = tl-tθ. Even if a spectrum is always completely read out, one can assign the readout time D ^ _nom ^ / k to a partial spectrum. The sought D _t preferably has approximately the size of D _t _ _nom - The distance which the trajectory travels during the time D _t is denoted by D. The following applies: D = D _t * V_Machine.

A burst measurement starts at a time to and has n partial measurements which start at the times to, ti, t ₂ ,..., T _n . The measurement is performed as if the section length were at U '= UD. Burst measurement over the reduced D section length consists of b partial measurements. The measurement time for the reduced section length is T '= U' / V_Machine. The measuring time for a partial measurement is tm = U7 (V * b). Since the measurement time is limited downwards by the readout time, tm> ts. A spectrometer has a minimum measuring time t _m j _n . It always applies

Obviously, W _n can not be smaller than ts. It also applies tm = (TD _t ) / b.

In the following, S (i) is the spectrum whose read-out process begins at a time ti. After reading out, the S (i) is divided into k parts: S (i) = [S (i, l), ..., S (i, k)] S (i, l) contains the light which is in Time interval [tj. _l5 tj] has fallen on the sensor. S (i, 2) contains light from the time [tj.i + D _t , tj + D _t ], etc

Example for the decomposition of a spectrum: S (i) is a spectrum that contains a wavelength range from 380nm to 980nm. S (i) is decomposed into k = 5 parts. S (i, l) has a wavelength range from 380nm to 500nm, S (i, 2) from 501nm to 620nm, S (i, 3) from 621nm to 740nm, S (i, 4) from 741nm to 860nm, S (i , 5) from 861nm to 980nm. For example, S (i) has 180 nodes, then the partial spectra have 36 nodes each. If the number of nodes is not divisible by k, not all sub-spectra have the same number of nodes. It is possible to reassemble the partial spectra to obtain a set of new spectra S '(i): S' (i) = [S (i, l), S (i + b, 2), ..., S ( i + (kl) * b, k)]. In order to obtain a complete new spectrum S '(i) = [S' (i, l), ..., S '(i, k)], pressure cycles are measured over at least k. The new spectra S '(i) have a negligible "spectral smear".

In the previous considerations, the partial spectra S (1, 2), ..., S (l, k), S (2,3), ..., S (k, 3), ..., S (k , kl), which are at the beginning of a burst, can not be assigned to S '(i). Also, the sub-spectra S (k, n-k + 1), S (k, n-k + 2), S (kl, n-k + 2), ..., S (l, n), in the end a bust measurement are included are not evaluated.

FIG. 2

The longer a measurement takes, the more useful partial spectra are present.

FIG. 3

Preferably, the measurement is performed such that each partial spectrum of a measured spectrum S can be used to be used as part of a new spectrum S '. This would be the case if the sub-spectra at the beginning and end of a burst could be assigned to each other. For this purpose, a "cyclic" measurement situation would have to be achieved, ie the image locations of the last k partial measurements would have to correspond-at least with the accuracy D / 2-to the image locations of the first k partial measurements shifted by D.

The measurement data, which are not needed to form this first spectrum, can therefore be used, if one does not stop the measurement after k pressure cycles, but measures a little longer.

If, in fact, measurements are made over more than k printing cycles, a new set of spectra, corresponding to an image area shifted by D, is obtained with each further revolution. If one defines z = tm / D _t (ie a partial burst has the length tm = z * D _t ), then you can use (almost) every sub-spectrum of the measurement data to compose a corrected spectrum. The assignment of a spectrum S '(i) to an image location has an accuracy of ± D / 2.

The burst preferably has a length of n = z * b + 1 partial measurements.

Although the measuring time at the respective measuring positions is much longer than before, there are two advantages:

One can do statistics: From each of the measured values, the densities of the colors that are present at the respective image locations can be determined. The density calculation for a color is all the more accurate the higher its area coverage in the image section is. When determining the density, one can increase the accuracy by a corresponding weighted averaging.

The relative measuring time increases, the relative positioning time decreases. The positioning of the measuring head takes a certain time, e.g. 0.5s. The measurement duration for a print cycle is on the order of 0.1s. If measurements are taken over one printing cycle and then a new measuring position is approached, the measuring rate is less than 20%. The system takes most of the time to position the probe. The longer the measuring head measures in one place, the less time is lost for positioning.

Claims

claims

1. A method for measuring optical spectra in web-fed printing, characterized in that a continuous measurement of the printed web takes place.

2. The method according to claim 1, characterized in that the measurement consists of partial measurements, which take place over several periods of a printing cycle.

3. A method for measuring optical spectra in web-fed printing, preferably in accordance with one of the preceding claims, in which a) from a printed, continuously required web with a section length U measured in the conveying direction with a spectrometer by means of burst measurement periodically at times tj optical spectra S (i) b) wherein a burst of each of the burst measurements consists of at least b partial measurements, c) the burst measurements are each performed for a reduced section length U 'which is smaller by a constant length D than the section length U, so that the track is between each other each of the reduced section length U 'travels corresponding distance, d) the recorded spectra S (i) are respectively decomposed into k partial spectra S (i, k) with k> 1, e) and the partial spectra S (i, k ) from burst measurements made at different times tj, preferably of temporal succession following bursts, to be assembled into new spectra S '(i).

4. Method according to the preceding claim, characterized in that the

Partial spectra S (i, k) to the new spectra S '(i) = [S (i, 1), S (i + b, 2), S (i + (k-l) * b, k)] are composed.

5. The method according to any one of the preceding claims, characterized in that the burst measurements, the decomposition of the spectra S (i) and the composition of the new spectra S '(i) over at least k pressure cycles are repeated.

6. The method according to any one of the preceding claims, characterized in that D <U '/ b, preferably D = U7 (b * k).

Method according to one of the preceding claims, characterized in that the spectrometer has a readout time ts per partial measurement and D is of the order of magnitude of the path traveled by the web within the time ts / k, preferably D equal to or less than (ts / k) * V, where V is the speed of the web.

8. The method according to any one of the preceding claims and at least one of the following features: the bursts each consist of the same number of partial measurements; the partial measurements are carried out at equal intervals.

9. Method according to one of the preceding claims, characterized in that each partial spectrum S (i, k) of a measured spectrum S (i) is used as part of a new spectrum S '(i).

10. The method according to the preceding claim, characterized in that the partial spectra S (i, k) at the beginning and at the end of the burst can be assigned to each other.