Offset-lithographic printing is essentially a binary process where areas of colour are applied as a checkerboard pattern (Half-tone). Dots per unit area rather than ink density determines the final printed colour. Traditional half-tone screens, which were physical overlays used in a photographic process to create a grid of dots of a particular colour (“separation”), have long been replaced by equivalent output from reprographic scanners.
To produce a full colour image, four separations are needed, three primary colours; cyan, magenta, yellow plus black. The black separation is used to compensate for process errors and the non-ideal spectrum of the coloured inks. Screen ruling, typically 0.05mm to 0.5mm, and screen angle are two main parameters which characterise a half-tone separation. Each separation represents a spectral band of the original image, which when overprinted reproduces the original image in full colour.
The original image, which may be a transparency or photograph, is processed by the “analyse” part of the reprograhic system, point by point (analyse spot size is 0.01mm dia. typically). The analyser extracts colour and spatial information from the original, which after signal processing, is split into its spectral colours i.e. Separations. The colours are chosen to correspond to the printing ink. Images may also be made negative, reduced or enlarged as required.
Image information is transformed by the E.D.G. (Electronic Dot Generation) into a half-tone image. The E.D.G. replaces the overlay, and superimposes a two-dimensional function onto the image information.
The image resolution corresponds to the smallest pixel size that can be generated on the film or medium. Normally this is not limited by the E.D.G. but by the electro-optic system which switches the data. Screen ruling (equivalent to D.P.I. in desktop publishing) corresponds to the repeat pitch of the half-tone grid. Generally, high resolution is required only with finer rulings. Larger rulings give a much coarser image and are used for lower quality printing (e.g. Newspaper images).
Although the trend towards C.T.P. (Computer to plate) systems has begun, most expose engines still produce the final image on film. These images are transferred onto the photosensitive printing plate, by contact printing, as a 1 to 1 image. At some point within the whole process (e.g. by controls in the expose scanner) account is taken of any positive to negative (black to white) reversal which occurs after contact printing.
Optics of Scanners
As sources of pure and high intensity light, Lasers are the predominant light source used in reprographic scanners. The wavelength sensitivity of the recording medium (film, plate) normally determines the laser chosen. With the abundance of different media on the market, performance comparisons are not straightforward. Laser cost per watt, Versus expose speed, versus media unit cost, versus cycle time are major considerations. Another complication is that more expensive plates generally allow longer print runs . Design cannot begin until a cost/benefit analysis has been undertaken to resolve these trade-offs.
For silverhalide based films Argon ion and Helium neon are the main laser types. Both types have excellent beam uniformity which allows simple and diffraction limited optical designs. The more powerful argon-ion lasers are bulkier and need robust mounting and alignment fixtures. Low power HeNe’s come in compact forms such as cylindrical and can be incoporated into systems with simple fixtures. Argon-ion lasers have been favoured because of their high photonic energy, high quality gaussian beams (TEMoo mode) and high resolution imaging.
Laser diodes emit at infrared wavelenghts of 780nm+, and have made a large impact on desktop printers. Their use in reprographics is more limited primarily because of poor beam quality, increased noise and restricted power levels. As a less mature technology there is scope for performance inprovement in the future, especially since they offer some major advantages; including direct intensity modulation, compactness and output at a desirable wavelength. Coupling a number of lasers on the same chip allows much higher output power, but has the disadvantage of larger beam size and unwanted radiation into sidelobes. Elements in this type of integrated array do not normally emit independently, unlike the distributed arrays discussed below.
Expose systems have also been realised using distributed laser sources, such as diode arrays. Each source in the array exposes a strip at the image (film) plane, and neighbouring strops precisely butt-up to each other. Full widths of film are exposed by a step and repeat method. This arrangement allows use of low power diodes or alternatively, enables improved beam quality by use of apertures to block unwanted sidelobes emitted from standard diodes.
Expose systems employing L.C.D. image bars work in a similar way, over much smaller dimensions. Each pixel of the L.C.D. matrix acts as an individual shutter. A typical image bar geometry spans several thousand pixels in its long axis (this is corresponds to the fast axis of an opto-mechanical scanner) and has a width of ten’s of pixels in the short axis. In this translation (slow) axis, the pixels expose overlapping strips to allow high resolution imaging over the whole length of the image bar. The illumination source can be simply a bank of high power L.E.D.’s.
Array expose systems are attractive because of the reduced power requirement of the individual sources and because they reduce the number of mechanical components.
Diode Pumped Lasers
Frequency doubled and diode pumped Yag lasers are also expected to have a major impact on C.T.P. (computer to plate ) technology particularly for printing on plates at the near infrared and thermal end of the radiation spectrum. Such lasers have peak power and near diffraction-limited quality at wavelengths of 1064nm. These beam qualities also lend themselves to efficient frequency doubling,tripling and quadrupling using non-linear crystals which can provide other potentially useful output wavelengths. Yag lasers with powers of several watts are now readily available.
Optics of Gaussian Beams
Pixel placement linearity (in x,y coordinates) and pixel resolution are primary determinants of half-tone image quality. When available laser power exceeds the film/plate sensitivity requirements, the cycle time is limited by the video bandwidth. Cycle time is defined as the time taken to expose an image over a given format, typically 600 X 600mm.
Designing for a smaller spot size to get higher resolution is not a fundamental limit in itself, however, other aspects of system performance become increasingly critical with smaller spot size. Depth of focus reduces, making mechanical positioning and flatness of the expose medium much more critical. The numerical aperture and thus size of optics also increases. The reduction in raster line width by half for example, means that cycle time is approximately increased four-fold .
When the gaussian beam diameter (1/e2 points) illuminating the objective is less than 4 times the entrance pupil diameter, the beam is self limiting and the Spot diameter do, is given by :
do = (4/pi) * (W * f/di)
Where pi = 22/7, W = wavelength, di is the (1/e2) diameter and f is lens focal length.
When no image data is applied, the exposure profile on the medium is simply that obtained from overlapping raster lines having gaussian intensity profiles. A design compromise is required between minimum overlap (desirable for faster cycle time) and exposure uniformity. Practical results show that exposure “ripple” is minimised when adjacent lines overlap at the 50% intensity points. This is the Fwhm, Full Width Half Maximum point, which is related to the 1/e2 diameter(do), by:
Fwhm = 0.5887 * do
In this case, system resolution r, is defined as:
r = 1/fwhm
Exposure ripple is visible as “staircasing” when screen angle is oblique to the raster line direction.
Depth of focus z, is defined as:
z = pi * do2/4V
This relationship confirms the trade off between small spot size and depth of focus. All these parameters relate to an unmodulated beam, which cannot impart any information onto photosensitive medium. In the raster (fast scan) direction minimum pixel size is normally limited by the modulator bandwidth (Acousto-optical and Electro-optical modulators are typical devices used). However, as in all laser scanners, the linear velocity of the spot (relative to the medium) combined with the characteristics and non-linearities of the medium, increase the effective spot size.
Some Major Scanner Types
Drum scanners have been used extensively as reprographic expose units. Optically they offer the simple advantage of on-axis imaging at a fixed distance between objective and film. Normally, such on-axis imaging is diffraction limited. Optical difficulties such as curved fields, speed changes and magnification variations, along the scan line are minimal.
Mechanical translation, combined with rotation of the imaging spot (expose optics) is needed to expose over the required format (see drum scanner schematic above). The drum on which the film/media is placed may be internal as above, or external (which requires drum rotation). The uniform movement (or rotation) of a heavy housings by mechanical means is the main disadvantage of drum scanners. As shown above, the whole optics housing is rotated to produce the raster line. Alternatively by rotating the mirror only (shown in green), on an air bearing, much higher speeds can be achieved. Since any astigmatism in the beam is also rotated by the mirror, spot errors can vary along the raster line, which is a disadvantage.
The other disadvantages of drum scanners are:
The film or media has to be curved to conform to the drum, which excludes the use of inflexible media.
Complex media transport mechanisms are required.
Precision engineering is needed to ensure that the axis of the drum and the translational axis of the optics are coincident and collinear (within limits).
As with all scanner types, the design, particularly in terms of depth of focus, must also cater for drum waviness and film thicknes variations.
In some systems, where the image spot is subdivided into smaller spots, telecentric imaging has been used to ensure system insensitivity to film plane deviations. Telecentric imaging is a widely applied technique, whereby the chief rays of the output beam(s) are maintained parallel to the optical axis, normal to the film plane. Thus film plane deviations have a reduced effect on spot magnification.
Flat Field Flat Bed Systems
The linear scan line (fast axis) in flatbed reprographic systems is generally created by opto-mechanical means (alternatives such as diode array and image-bar type systems are ignored here), however, as in drum scanners, the translation in the slow direction is always done mechanically.
The primary task of the scanning optics is to convert an angular deflection, into a linear velocity. The ideal characteristic takes the form:
dx = f * da
Where a constant angular change da, is transformed to a constant linear change dx.
In practice, lens characteristics take the form:
dx = f * Tan(da)
So that equal angular increments result in larger linear increments, at the edge of the scan. This results in positive “pincushion” distortion which can be optically compensated by introducing negative distortion in the lens design.
Without auxiliary optics, even a compensated lens does not produce telecentric imaging. This is because the principle rays vary in angle, relative to the film plane, at scan line. Magnification errors caused by variations in film plane position will thus vary across the scan line. Toroidal lenses can be used to improve the tolerance to such film plane errors.
Other pixel placement errors (jitter) occur whenever there is a synchronisation error between the X,Y image spot coordinates in space and the electronically stored pixel coordinates. Speed variations in the polygon and optical errors in the facets are primary causes of mismatch. Using start/end of scan sensors, certain jitter errors can be corrected along the scan line by adjustments to the data clock frequency. Typically, residual pixel placement errors of +/- 1% are acceptable.
The design of the imaging objective must take into account spot size and scan linearity requirements and other factors such as available space etc.
In pre-objective scanners, the scanning element (generally a polygon) scans the beam over an angle a, into the imaging objective, which images over a straight scan-line field L . Because the pre-objective lens must provide a well corrected image over the whole scan line, it requires a complex design. Note that the optical scan angle a, is twice the mechanical scan angle. When these objectives are used with collimated input beams, which is normally the case, spot size is limited only by the lens aperture and back focus distance. Again auxiliary optics are needed for telecentric imaging.
In post-objective scanners, the polygon is placed after the imaging objective. Because imaging is always on axis, the design of a post-objective lens is relatively simple. Satisfactory optical performance can be obtained with a four element lens. Since the mirror (polygon facet) pivot point is at a constant distance to the focused image spot, the swept field describes a circular arc. The highly convergent beams, which occur in high resolution systems, can make it difficult to use post-objective scanning. This is because the mirror folding of highly convergent becomes increasingly difficult and results in geometric or space problems.
Using a novel arrangement of hyperbolic and parabolic mirrors as auxiliary optics Wim Van Amstel, has shown it is possible to obtain a flat field and telecentric imaging over an extremely wide format.