Undergraduate Thesis: Design and Construction of a Prototype Image Setter

Achievement of Excellence Award

Note: This thesis was done for Universidad Iberoamericana in Mexico City so some images have legends is Spanish, and the full report is also in Spanish.

Download Complete Report [4MB PDF] (in Spanish)

An imagesetter is an industrial version of a laser printer but instead of creating final prints directly it is used to create a printing plate or negative that is then processed and mounted on a press to produce thousands of copies.

The goal of this project was to design and build a prototype imagesetter capable of directly engraving ceramic rollers with a high power laser, thus eliminating the need for negatives, etching and mounting. My part consisted of building a low-budget prototype of the design.

Main Idea The design requirements called for a machine capable of accepting a variety of roller diameters and lengths. Since the rollers to be used are heavy (up to 300kg) it was decided to rotate them at constant speed and scan them by columns, instead of rows like normal laser printers and most imagesetters. Also, to assure quality and avoid dot deformations as well as variable resolution, it was decided to move the laser source or at least a mirror axially instead of scanning with a rotating mirror.

The final mechanical design was planned to be much different (to accommodate different diameter and length rollers) but the prototype, that captured the essence, was like that shown on the figure to the left. A DC motor with its gearbox was used to rotate the roller which was coupled to an encoder and controlled for constant angular speed through a closed-feedback loop. A stepper motor coupled directly to a ballscrew was used to position a laser-pointer along the roller's axis of rotation. The stepper motor operated on an open-loop with only a home switch as feedback.

click to enlarge

Electronics & Operating System Rather than using a single microprocessor as a CPU a distributed architecture was chosen to simplify debugging and maintenance as well as to reduce cost. Each actuator (DC motor, stepper motor and laser pointer) has its own Atmel-AVR-RISC microcontroller and driver electronics running asynchronously from one another. The laser controller was also responsible for serial (RS-422) communication with the host computer and asynchronous SPI communication with the other microcontrolers. All three controllers are configured by the host computer but their operation is regulated directly by the signals from the custom (because of budget constraint) encoder. It is worth noting that the entire OS was programmed in assembly language completely based on the chips interrupts, all the controller's main loops were always idle. Since all systems relied on the signals from the encoder to regulate their operation this architecture proved to be very robust and reactive. You could hold the roller on the prototype to prevent it from spinning without causing printing disruptions.

click to enlarge

The Prototype Once the electronic architecture was finished a simple wooden mock-up was built to illustrate the operation principles and overall functionality. This model was not intended to actually work, however I decided to take it into a darkroom, cover the roller with photographic paper and test it. With a little debugging the model was printing within no time. The main drawback was that the laser-pointer was transmitting at the wavelength that B&W photographic paper is least sensitive to (Red) so a lot of exposure was necessary and thus printing took some time. The wooden model successfully proved that all the systems were working and it was not necessary to build a better model.

Thermal Analysis Since the final goal was to use a high-power laser to carve tiny holes on a ceramic surface a thermal analysis of the material exposed to an energy beam was conducted. Standard FEA tools did not support sublimation or material removal so it was necessary to program a simulation.The movie shows the thermal response of Aluminum Oxide (Al2O3) with a sublimation temperature around 3000°C to a 300microsecond laser pulse of 25W focused on 60micrometers, and its cooling. Experiments conducted in the Instruments Lab at UNAM with a pulsed Argon Laser of similar ratings resembled the simulation.

Optics Design During design, the alternative of changing the laser-beam diameter to directly produce halftone cells in one pulse instead of rendering them from several pulses was considered. The principal constraint was speed, since commercial CO2 lasers were capable of firing at 10kHz (important to reduce the overall imagesetting process) the optics had less than 100microseconds to configure themselves. The main features of the design outcome are that the moving lenses have to move only a few micrometers to produce a significant variation on the beam diameter, and both the input and output beams are collimated. It is worth noting that this design would not have worked for a high power laser outside a vacuum because the beam crossings concentrate so much energy that air literally explodes, however it is still a good design for lower power lasers.