• Font Size:
  • A
  • A
  • A

Feature Articles

Machine Vision Adapts to the Solar Industry

by Winn Hardin, Contributing Editor - AIA

Wafer BeispieWhen the semiconductor industry transitioned to 300 mm wafers several years ago, larger, more fragile wafers opened up new business opportunities for machine vision suppliers and integrators. In today’s solar energy industry, similar drivers are again expanding the role of machine vision in modern manufacturing. 

As semiconductor wafers grew from 200 mm to 300 mm, these thin pieces of multi-layer crystal became more susceptible to breaking, chipping and other defects caused by moving the product from one production machine to the next. At the same time, larger wafers meant that each wafer cost more to make but potentially generated more revenue. So each 300 mm damaged or broken wafer became a greater loss to the manufacturer than the loss of a similar 200 mm wafer. 

As a result, manufacturers made greater use of automation for manufacturing and transferring wafers from one cell to the next, along with new data matrix and other identification techniques that allowed the automated handling equipment to track each wafer through the production process. With this automated information gathering, semiconductor manufacturers were better able to protect valuable product, while collecting manufacturing equipment performance information. This information allowed them to improve yields and further increase the semiconductor manufacturing process, thus improving productivity and profitability. 

Solar cells face many of the same challenges facing the integrated circuits industry. Like wafers that eventually become integrated circuits, solar cells are manufactured by applying very thin layers of materials on fragile crystalline or glass surfaces. And like integrated circuit manufacturers, solar cell makers depend on automated manufacturing and data collection as they quest for parity per megawatt compared to traditional energy sources, such as gas and coal-based electrical plants. 

Three Opportunities Under the Sun

Solar cell and module manufacturers use machine vision systems for three general purposes: inspecting the products, identifying and tracking the products, and assembling the products, typically as a guidance system for a robot. 

The requirements for a machine vision system changes depending on whether the solar cell is based on crystal silicon or thin-film glass, according to Brian McMorris, Emerging Markets Manager at SICK (Minneapolis, Minnesota). “The machine vision processes required for crystalline wafers are similar to semiconductor wafer applications for inspection and vision. The wafer is inspected for physical defects such as edge chipping, bowing or ‘potato chipping,’ cracks, scratches, saw marks and pits. The area to be inspected is generally less than 20cm x 20cm compared to the thin-film glass [solar] panel, which is much larger - as much as 2m x 2m. The surface of the crystalline wafer is more critical than the thin-film glass panel because the crystalline wafer is active and forms the circuitry of the solar semiconductor, whereas the thin-film panel is passive and provides an insulative base on which the active materials are deposited. So, the inspection of the glass panel is more coarse than for the crystalline wafer - millimeter rather than micron [spatial] resolution. 

“The identification task for both crystalline wafer and thin-film panel is similar,” continued McMorris. “In both cases, the target is a very reflective material making optical techniques for identification, like bar or matrix code, quite challenging. In both cases, it is undesirable and/or difficult to place or print a high-contrast code on the material. The preferred means of coding is laser etching. This suggests a 3D technique might prove more desirable than 2D in some applications.”

Once the silicon crystal boule has been cut and patterned for converting light energy into electrical energy, or the glass panel has been overlaid with various semiconducting and insulative materials for the same energy conversion purpose, it is time to test the product, classify its output for grouping with other panels to meet final customer requirements, and then assemble the final module or panel for shipping and installation. With the help of machine vision, robots handle most of the delicate active materials with a speed and dexterity far beyond that of humans. 

Go Robot Go

To achieve parity with traditional energy sources, solar panels have to become cheap and robust. A few ways to help the industry achieve its price points are to produce solar panels faster, and with fewer defects. That means automation, and in this case, robotics. 

“Developing a machine vision system for high-precision solar panel positioning has several challenges, including the imaging over large areas and in conditions where the available light is not homogenous and changes throughout the shift,” explains Endre Toth, Director of Business Development at Vision Components GmbH (Hudson, New Hampshire). “The lighting randomly changes, and you have to accurately and quickly find a square-shaped tile in the camera’s field of view. But the tile may have rounded edges or other challenges. You have to find the center and rotation and orientation very accurately to be able to place the panel in the module or cell to maintain production yields.”

“Thanks to improvements in computational power, smart machine vision cameras can now use complex image processing algorithms to locate a point’s physical position to within 100th of a pixel, which also reduces the hardware costs because you are doing most of the work with math,” Toth adds. “These days, the limitations in positioning are typically from the robot and the mechanical systems, not the vision system’s ability to locate a precise point. Cost considerations are also a big deal for solar manufacturers, which is why smart cameras are so attractive, because you have the camera, frame grabber, memory controller, and interfaces integrated into small compact unit. Our VC4466 actually includes a direct monitor output on the camera as well. All this integration means that the system integrators do not have to put in a lot of cables or different boxes into a system, which also add to the acquisition, operation and maintenance costs of the final system. With a smart camera, you can swap one single unit to fix it. While these considerations may seem mundane, they are important selling features to operations that need to stay up and running without pause. Usually, our prices are compared to a PC host unit with low cost camera, but when you have to integrate the whole solution together, it turns out the that real cost of ownership is very different than what most PC host suppliers will quote.” 

Solar’s Bright Appeal

With the recent drop in oil prices, many are questioning the ongoing appeal of alternative energies, however many pundits believe that its only a matter of time - perhaps a year - before oil prices return to their $100+ per barrel levels. As prices of oil and gas rise, solar energy will continue to gain pricing momentum. 

“The solar industry is driven by the goal of ‘parity cost’ with grid electrical sources from fossil fuels, nuclear and hydro,” explains SICK’s McMorris. “To the extent parity can be achieved at current electricity production costs, government subsidies are unnecessary and higher oil costs are unimportant to the industry's success. To achieve this goal requires better techniques for fabrication and more efficient and consistent production with less waste material. Tracking [solar] units through production helps to achieve quality improvements within a given process technology. Inspection and identification are essential to any quality control methodology. There will be a continuous refinement of solar photovoltaic production techniques that will yield increasingly more efficient and lower cost panels. Thin film has many production advantages over crystalline wafer in that the entire panel is manufactured as a continuous process, but thin-film offers only one-half the solar electric conversion efficiency as the various crystalline wafer designs. For broad solar segments [crystalline and thin-film], inspection and identification will help engineers achieve better performance over time.”



There are currently no comments for this article.

Leave a Comment:

All fields are required, but only your name and comment will be visible (email addresses are kept confidential). Comments are moderated and will not appear immediately. Please no link dropping, no keywords or domains as names; do not spam, and please do not advertise.

First Name: *
Last Name: *
Your Email: *
Your Comment:
Please check the box below and respond as instructed.

Search AIA:

Browse by Products:

Browse by Company Type: