1 February 2002
Rebuilding a Laser Cutter System
Musa K. Jouaneh, Ph.D., David B. Butler, J.T. Woodhouse III
Integrating a machine rebuild project is challenging, especially when your target system is maintaining its production schedule.
Rebuilding a laser system used for cutting industrial fabric and laminates involved making both hardware and software changes. The project required us to replace all existing actuators, as well as the control, operating, and laser systems, on the current machine. What follows is a discussion of both the integration issues we faced and the lessons we learned while maintaining continuous production.
The 1983 Lectra Systèmes Model E92 continuous flow laser machine is a gantry-type cutting device (Figure 1). Its laser head is mounted on an xy positioning carriage. Cloth is fed from a powered take-out spool, preplaced under the head on a tractorlike bed, and held in place using vacuum. The bed advances the material and stops. Original machine control was via a bank of PDP-11–like boxes. The laser system cuts by using a 5/8-inch-diameter beam directed by five angled mirrors up into a collimating lens. The lens is mounted in a flying head (Figure 2) positioned over the material to be cut. A plot file of the geometry to be cut is dumped to the xy servo controller, which directs the flying head to cut the batch held stationary on the bed by vacuum. The machine had five different actuators, one DC servomotor to control the bed, two servo-controlled actuators for the head, a stepper motor to position the cloth crosswise, and a DC servomotor to control the cloth feeder.
The controller, a 20+-year-old system no longer supported by the original manufacturer, was based on erasable programmable read-only memory (EPROM) that wasn’t backed up. Worse, there’s no documentation for potential service assistance, which made the existing machine a potential disaster in a production schedule. In addition, the existing control system had many limitations (e.g., restrictions on data file names and difficulty both in software control of laser intensity and in performing rework operations). For these reasons, the decision to rebuild this system—making it state of the art in computer-controlled cutting—wasn’t difficult. To provide flexibility, we discarded proprietary control systems in favor of a new design based on existing off-the-shelf basic components.
Table 1 compares the overhauled system’s new major components with those in the old system. With the exception of the frame and the drive system, we replaced almost every part on the machine with a new one.
Figure 3 shows a block diagram of the new control system. The system physically consists of a Pentium II PC interfaced with a Warner Electric eight-axis controller (MX2000) using a serial line and a 32-channel isolated digital I/O line.
The PC acts as host, sending commands to the Warner controller and displaying status information for the user. All commands are sent using the optically isolated I/O lines. Data (position information for the xy servos, plus laser intensity and machine setting information) is transmitted via the RS-232 serial line and is stored in 32 kilobytes of nonvolatile memory.
Due to this control system change, we found it necessary to develop software to translate the plotting files into a form that could be sent to the Warner controller. These plotting files, written in a Hewlett-Packard graphics language format, are generated by several software packages, including PlotterPilot and Pattern Smith from Calif.-based Autometrix. We designed the translation software to accept both formats.
The control software residing on the Warner controller was structured as four separate, continuously scanned tasks. These were set to control, respectively, the xy servo system, the bed axis, the laser, and the feeder system. We structured the code in each task as nonblocking state transition logic, where in each scan only the code in one state was executed. This allowed fast code operation, giving a seemingly parallel task operation.
The operator interacts with the new machine via a PC-based Windows application. This application, developed using Visual Basic 6, allows the operator to perform the following functions:
|Command the bed axis individually|
|Command the bed axis and the feeder simultaneously|
|Monitor the machine status|
|Monitor any errors sent by the Warner controller|
|Read a plot file|
|Set the speed and laser intensity|
|Set the starting and ending bed numbers|
|Start/Stop the cutting|
|Transfer plotting data to the Warner controller|
After the user selects a plot file, the software reads and analyzes it. The software then informs the user of the number of table (or bed) moves and the total number of data points the specified plot file has. When the operator issues a command to start cutting, the software downloads the next bed data to the controller and informs the user of the current table (or bed) number stored in the controller. This information is updated every time a new table data is transmitted to the controller.
The interface allows the user to perform cutting on a bed-by-bed basis, and we added controls to specify the starting and ending beds. This feature is important because it allows the operator to cut only a select number of beds for rework situations.
The new software also allows two different laser settings: one for cutting and the other for marking (drawing numbers, or marks, on cloth). The analog voltage input control signal to the laser controller is modified in real time to allow for this functionality. Using marking (slight scorching) and cutting settings for the laser eliminates the use of a pen-based marking system.
A major constraint placed on us during system development was keeping the old system running until a complete changeover was ready. Consequently, we had to perform the development in segments. Table 2 lists the sequence of events in our rebuild operation.
We started by developing code to translate the plot files into a form that could be sent to the Warner controller. Developing the user interface software followed. Then we combined these two software modules into a single program capable of reading a plot file and sending commands to the Warner controller to control the xy servos and the bed axis. We performed these activities without disturbing the existing system. We tested the developed code’s functionality by performing tests on the new actuators without mounting them on the machine.
Afterward, we started testing the new system actuators on the machine. We were required to accomplish trials on weekends using the new system actuators and to change back quickly to the original components for production. Our solution was to fit the Superior servos with mounts that exactly fit the French metric foundations. Change out, a process we employed successfully over a three-month period, took four hours.
One month into the conversion, the EPROM-driven bed controller and bed servo failed. We then installed the new Superior bed servo and ran bed (table) moves hybrid from the Pentium II driving the Warner MX2000 bed-axis controller, while running the rest of the machine from the original components.
Following this, we began testing the new laser system. Once we’d verified that the laser, the new bed and head servomotors, and the controller were working, we shut the machine down for a week to integrate the actuators with the new laser and control systems. Afterwards, the rebuilt machine underwent a one-month period of slow production at slow speed, with refinements/additions to the new control system. Our final task was to integrate the feeder control system.
The new control system is a significant improvement over its predecessor in terms of ease of use and functionality. Below is a list of some of the features we implemented in the new control system.
|Ability to: |
Control cutting on a bed-by-bed basis
|Perform discrete manual testing of the laser system|
|Control the bed axis individually or with the cloth feeder system|
|Online control of laser intensity to allow for marking or cutting|
|Use of a standard generic industrial controller and software|
In addition to the above-listed improved features of the new control system, our machine rebuild operation resulted in a more efficient light path for the laser system. As a result, much less power is used to perform comparable cutting operations.
We faced several difficulties in the course of our rebuild effort. One challenge was to proceed without shutting down the old system. This created some logistics problems that we addressed by planning our testing around the production schedule.
Another hurdle was the laser head’s unique drive system. As designed, an 80-foot-long timing belt (shown in Figure 2) is wrapped around involute sheaves in a figure H configuration. Two actuators, mounted at the top of the “H,” control the head’s xy motion. When each of these actuators is commanded separately, the head moves at 45° relative to the Cartesian xy system. Thus, producing motion along the x or y axis requires synchronizing the two actuators’ rotation. This H-drive system required us to develop a custom homing routine, as homing the xy head along the Cartesian x or y directions requires moving the two head motors simultaneously. In addition, all of the xy head’s computed displacements needed to first be resolved into the corresponding actuator motions before commanding those actuators.
Furthermore, the Warner controller has a limitation in processing coordinated motion moves: The Warner system cannot round sharply cornered segments. Consequently, a path with a sharp corner will experience overshoot motion around that corner. To circumvent this, all the xy motion head data is preanalyzed to determine all sharp corner locations. At each occurrence, the original motion path is split into two smooth paths (with a momentary stop between them) joined at the sharp corner.
The new machine, which was successfully rebuilt, is being used in production. MC
|The authors wish to thank Richard Morlock of Machine Parts Corp., Providence, R.I., for his support during this project.|
Musa Jouaneh received his Mechanical Engineering degrees from the Universities of Louisiana (B.S., '84) and California at Berkeley (M.S., '86, and Ph.D., '89). He is an associate professor in the Department of Mechanical Engineering & Applied Mechanics. His research interests include automation, motion control, and design of high-precision positioning systems. Contact him at Wales Hall, University of Rhode Island, Kingston, RI 02881; tel: (401) 874-2349; fax: (401) 874-2355.
David B. Butler performed R&D in the process instrumentation and nonwoven fabric industries for 8 and 12 years, respectively. He has extensive patent experience in product development. He founded the Development Laboratory at the University of Rhode Island in 1980 and the Deep Ocean Simulation Facility in 1984. David is president of Applied Imagination, Inc. Contact him at 130 Boston Neck Road, North Kingstown, RI 02852; tel: (401) 294-9622; fax: (401) 294-7048.
J. Tim Woodhouse III is president of Hood Sailmakers, Inc. He began making iceboat sails while still in high school. He won the world championship and was three-time North American Champion with a sail of his design. In 1974, Tim opened his own business in Michigan and manufactured sails for yachts in the Great Lakes area, winning many other titles with his designs. In 1979, he became affiliated with Hood Sailmakers and ran a licensed company under its trademark. After several successful years as a licensee, Tim organized a buyout of the parent company and moved to the East Coast. Contact him at 23 Johnnycake Hill, Middletown, RI 02842; tel: (401) 849-9400; fax: (401) 849-9700.
Return to Previous Page