As the U.S. weighs its options to reduce dependence on foreign oil, one technology is emerging as a clear winner: solar energy.
This decade has seen a boom in solar technology. In 2006, the solar industry raised more than $4.4 billion in the capital markets, a strong indication Wall Street believes the technology will play a vital role in the nation's energy mix. And as 2007 came to a close, the solar electric market was on track to grow by more than 60 percent versus the 2006 pace, according to the Solar Energy Industries Assn.
Moreover, if Congress approves a proposed 80-year extension of the 30 percent tax credit for business investments in solar, as well as tax breaks for home solar installations, the growth rate for solar will continue to soar, creating thousands of jobs, say solar experts.
This sunny outlook poses big challenges for companies that produce the photovoltaic wafers that are the building blocks of solar energy systems. California-based Owens Design Inc. (ODI), an engineering firm with 25 years of experience in developing automation systems for equipment companies in the semiconductor and data storage industries, was recently approached by a startup solar manufacturing company. The solar firm had successfully demonstrated their photovoltaic process with manual tools, but it clearly needed automated equipment to ramp up its capacity to meet rising industry demand.
Meeting the Challenge
ODI faced some tough obstacles in tackling this project. The PV process used by ODI's customer precisely separates fragile PV cells and sorts and places them into custom carriers for the next operation. Machine vision would also be needed at the input and output stages to inspect for surface and dimensional defects and provide positional feedback for robotics. The tool complexity and compressed development time also required a new approach to the control system, electrical design, wiring and software development.
For system control, ODI's solution featured a PC running Windows XP and a Mitsubishi PLC. The PC provides the user interface (GUI), an interface to the visions system and a connection to the system's PLC. The PLC provides the real-time motion control and sequencing operations. Figure 1, page S16, shows a block diagram of these controllers, as well as the other key components in the system.
ODI selected the Mitsubishi Q series PLC as the control platform because of its processing speed, small size and software capabilities. The PLC's main processor is the Mitsubishi Q06HCPU CPU, which handles all machine I/O programming in ladder logic. In addition, a motion processor, the Mitsubishi Q173HCPU, handles the system's 30 servo axes, using flow chart programming.
The Q series PLC allows operators to modularize the application into many separate programs. This reduces lead time by enabling several engineers to work on the PLC software simultaneously. Moreover, ODI could easily replicate the modular programs to support the multiple process modules in the automation tool. Further, the CPU allows multiple users access simultaneously, so troubleshooting and debugging tasks no longer depend on a single programmer.
Time and Money Savings
The automation system's motion CPU links to 30 MR-J3 servo amplifiers via a daisy-chained fiber-optic cable (SSCNET III) to provide high performance and reliability, while eliminating complex wiring harnesses. Home and limit sensors are wired directly to the MR-J3 amplifiers to reduce I/O requirements, creating additional savings.
This simple connection method greatly reduced wiring and system debug time, says ODI Senior Electrical Engineer Stephen Chu. We also saved time and improved reliability by eliminating the electrical noise that plagues traditional servo systems.
The motion CPU offers flow chart programming to simplify the coding for motion control. All servo amplifier parameters are sent from the motion CPU to eliminate the need to download parameters individually via computer. The motion CPU was easy to set up and enabled fast software duplication on this multiple module system, says Mike Ruble, the lead software engineer. Auto tuning of motors also worked well.
The system consists of multiple process modules operating in parallel. To save time and cost, the prototype system (hardware and software) include only one process module. The controls and camera selection enable rapid duplication of the software once the process module is finalized.
Essential Role of Vision
Machine vision is vital to the successful performance of the system. Five smart cameras inspect and locate the product at two stages: system input and process module outputs. A single camera at the input station inspects for surface and dimensional defects on the PV wafers and establishes the position of the parts. This data is transmitted to the PC and then passed on to the PLC. Cameras at each of the four process module outputs communicate directly to the PLC through Ethernet and indicate pass/fail status of processed parts. Mitsubishi provides the Ethernet link that connects the PC and PLC.
ODI selected the Iris P-Series camera from Matrox Electronic Systems for performance, system cost-effectiveness and ease of programming. This smart camera runs under Windows CE.NET. The camera application code is written on a PC and downloaded to the Iris over an Ethernet link. The modular architecture enabled ODI to use the same camera software for all four process modules. To further minimize software development time, the same camera and software were used in a tool upstream of this process module. All image processing takes place at the cameras to eliminate need for high-speed communications or processing at the PC.
Useful Design Tools
In developing the overall PV automation system, ODI engineers relied on Autodesk Inventor for 3-D modeling. To speed the design process, the system was divided into functional subsystems to allow several engineers to work on the system in parallel. A common top level assembly was updated regularly to allow the engineers to verify fits of each subsystem. Engineers found 3-D modeling was a great help in verifying the interaction of the many custom mechanisms and subsystems, while minimizing errors and reworks. Inventor also provided the ability to animate the complex motions to effectively communicate and facilitate decisions with the customer.
For electrical design, engineers used AutoCAD Electrical, again reducing time and errors. This CAD tool was particularly useful for a project of this size by automating repetitive tasks. The system has 67 pages of schematics. The I/O assignments for this system were imported directly into AutoCAD EE from an Excel spreadsheet. AutoCAD EE automatically generates the schematic and inserts the addresses and description text. The software also provides automated reports, such as bill of materials and From/To wire lists.
In the end, ODI succeeded in developing a fully automated PV processing tool that can produce more than 5,000 parts per hour, using several process modules operating in parallel. The system includes 30 servo motors, 15 dc motors, 22 pneumatic actuators, five smart cameras and more than 200 I/O.
ODI worked collaboratively with the customer to complete the prototype in 18 weeks. The prototype tool achieved the cycle time and yield goals the customer needed for production. What's more, the customer successfully secured its next level of funding and established long-term supply contracts. ODI has subsequently already manufactured a next generation of the tool with a modified process, further helping the customer meet the rapid run-up in demand for solar energy components.
Bob Fung is director of engineering for Owens Design Inc., Fremont, CA.