June 2002
Newsletter

• News and Events
• Understanding Optical Linear Encoder Error
• Calculating Vision Sensor Resolution
• Vision-Guided Motion

• Olympus Controls Automation Seminar in Seattle on July 25th at 8:00 am at the Holiday Inn in Renton, WA. Technical presentations discussing servos vs steppers, vision-guided motion, linear encoder technology and piezoelectric motors. Email Mary Jensen to reserve a seat.
• Paul Woodhouse joined Olympus Controls as an inside applications engineer.
• David Kennedy joined Olympus Controls as a Seattle area automation engineer.
• Parker Hannifin makes strategic partnership with Animatics Corporation to mount the patented SmartMotor drive and controls to the impressive BE series 8-pole servo motors.

Optical linear encoders can provide extremely fine positioning feedback. An optical tape scale can be applied in minutes to an axis and provide up to 10 nm resolution. Successfully pushing a linear encoder to its limits, requires an understanding of the errors involved.

1. The sub-divisional error band is a small cyclical error based on the encoder pitch. Typically +/- 0.15 microns.
2. The slope error is a property of the encoder tape, temperature and installation. A two-point slope correction will calibrate the scale and provide a total system accuracy of at least +/- 3.0 microns per meter.
3. The linearity error band is a result of the manufacturing process. High precision manufacturers like Renishaw promise +/- 0.75 microns in any 60 mm section while often delivering higher performance. A multiple-point error-mapping algorithm, using an interferometer, can significantly decrease the linearity error band.

More information on Renishaw encoders can be found at www.renishaw.com.

1.
2.
3.
4.

Machine Vision is reaching the masses. Engineers and designers can now incorporate digital, CCD-based vision sensors into their quality control, measurement and positioning applications. Using feature based algorithms, todays vision sensors are more flexible and more robust than ever before. This allows vision sensors to be using in new applications every day to increase quality, efficiency and machine capabilities.

With new technology, it is important understand the limitations. Here is how to quickly calculate the resolution of a vision sensor to determine if it will work in your application.

1. Determine the FOV (field of view) of the lens. (Often the object or feature length.)
2. Determine the CCD width in pixels (Usually 640 or 1280)
3. FOV / CCD width = Pixel Resolution.
4. Divide by 10 = Sub-pixel Resolution.

Example: Looking at a 2 inch hole with a 640x480 CCD, add 0.25" on each side for a total of 2.5". This would result in 2.5 / 480= 0.0052 inches per pixel. And adjusting for sub-pixel resolution 0.0052 / 10 = 0.00052 inches per pixel. (Most industrial vision applications only rely on 1/4 pixel resolutions 0.0052 / 4 = 0.0013 .)

Note: All vision sensors are light dependent, consistent sub-pixel calculations require controlled positioning and lighting of the object to be measured.

More information on DVT SmartSensors can be found at www.dvtsensors.com.

1.
2.
3.
4.
5.
6.

When the largest full-line motion control company (Parker Hannifin) and the leader in smart vision sensors sold (DVT) team up to provide vision-guided motion, the results are phenomenal. Process, Design and Manufacturing engineers can easily add vision-guided motion to their projects.

The information flows through the system as follows:

1. Capture an image.
2. Match the object with the expected position.
3. Determine the position and rotation from the expected position.
4. Scale and values from pixels to real values.
5. Send data to the controller (Ethernet).
6. Command motion using variables from the Smart Sensor.
7. Capture an image again for a closed loop.

Note: Smart vision sensors, like controllers, motors, drives, mechanics and encoders, have error associated with them. It is important in the mechanical design of the system to allow for the part to be located within a positional and angular tolerance. This tolerance is based on the variety and contrast of the parts being tracked. The best way to determine this tolerance is to load a series of part images into the Free DVT Frameworks emulator and log the variance in measurement data.

A white paper on vision-guided motion can be found at support_services_vision_guided_motion.htm.

More information on the Parker Compumotor 6K motion controller can be found at www.compumotor.com.

More information on DVT SmartSensors can be found at www.dvtsensors.com.