Oscillation Compensation in On-the-Fly Laser Processing via Optical Tracking Sensors

In today’s industrial solar cell production, laser processing is a standard technology and will remain important for future developments, such as silver-free solar cell concepts. High throughput laser processing of more than $10.000\,\text {wafers per hour}$ with a high precision of 10 to $100\, {\mathrm {\mu \text {m} }}$ is required. Current laser systems use complex automation and transport systems for accurate positioning. However, these systems are costly, large, and have limited throughput. On-the-fly laser processing, with wafers processed on a conveyor belt, offers a compact and cost-effective high-throughput approach. However, its accuracy is limited by transport imprecision. To overcome this, we propose a novel system using real-time measurement of speed fluctuations to adjust the beam position via feed-forward control to achieve both high throughput and accuracy. An FPGA-based array of optical tracking sensors was developed and used in an on-the-fly demonstrator setup. The sensor’s accuracy is characterized and demonstrated in a one-dimensional oscillation compensation. Oscillatory deviations of the laser markings are reduced from 82.2 to $13.2\, {\mathrm {\mu \text {m} }}$ ( $1\,\sigma $ ), a six-fold improvement. Thereby, sampling delay limits the control bandwidth. Offline analysis shows that Kalman filtering could compensate for this, and still reduce the estimation uncertainty by $23\,\%$ . Note to Practitioners—We have developed a compact and robust laser processing system that processes solar cells while they are moving. Instead of depending on high-precision transport systems, our system actively controls the laser beam’s position to correct for any speed fluctuations of the moving solar cells. The prototype created uses an array of optical tracking sensors that measure the speed of the solar cell on its surface. This speed information is used to generate an encoder signal for the laser scanner, which then adjusts the laser beam’s position in real-time. Advantageous are the high sampling rate of the sensors and the low latency in processing the measurements. This approach could also be adapted for other applications involving high throughput laser processing.