Many machine vision applications throw up challenges that can’t be met using standard components. Since its formation in 1999, Gardasoft has made a significant investment in R&D. The resulting expertise in LED system hardware and software for control, measurement, triggering, sequencing, programming and communications, puts the company in a unique position to design and manufacture bespoke controllers. Over the years it has designed a number of custom controllers for individual applications or for OEMs.
Running LEDs at low output intensity presents a major challenge because a small variation in light output can be proportionally very large. Just 10μA of current can cause an LED to illuminate.
Nevertheless, there are many machine vision applications where very low illumination levels are needed to avoid saturating the camera sensor. These include:
- Backlit applications where work pieces are being imaged in silhouette and the background illumination is bright
- Brightfield illumination of highly reflective objects, especially metals
- Transmission inspection of transparent objects
- Test equipment for high-sensitivity sensors
To operate at very low light output it is very important to control all light leakages and prevent the LED from illuminating at the wrong time. The control system needed to have very low noise and achieve an output that is stable within a few μA. This performance is orders of magnitude better than most electronics can achieve and gets close to the limits of physics. From a machine vision integration point of view, the controller also needed to have all of the attributes of a traditional lighting controller, with multiple output channels, fast accurate pulsing and safety features.
Gardasoft was able to overcome these challenges with a design that minimised the effect of noise, errors and crosstalk and achieved a stable, adjustable current drive from a maximum of 90 mA all the way down to just 7 μA. The exceptional current stability provided the highly repeatable lighting intensity which is essential for reproducible machine vision measurements.
A built-in miniature web server enabled it to be configured and controlled from a PC running an image processing application or directly by using simple string commands from an application program using TCP/IP or UDP. The unit also addresses the reliability of illumination for machine vision systems by continuously monitoring the health of the lighting. By monitoring the operating voltage of the LEDs, it can alert the user if there is a change that indicates the LED is nearing the end of life. This safeguard enables the user to schedule light replacement and avoid unexpected and expensive manufacturing downtime.
