Delta CS24D - Manual (Page 3)

CS24D Startup Guide
Page 3 of 4 Delta Computer Systems, Inc.
The next important selection is the Teach Criteria.
Color Only is the best choice for accommodating large variations
in sensor-to-target distance but it cannot be used if any of the
target colors are shades of the same color (e.g. light gray and dark
gray). Color plus Intensity handles shades of colors, and gives
equal weighting to the RGB readings and the Intensity or
“darkness” of the color.
The Error Limit parameter is also calculated for you by the
optimize routine, although the default value will also probably work
fine for a quick test. Averaging should be set as high as possible,
especially with dark colors. The Stretch Outputs parameter can be
left at zero for most applications.
The last Setup Wizard screen allows you to review the parameters.
Pressing the Finish button will download the parameters to the CS24D
color sensor.
That’s all there is to it! The CS24D can now differentiate between the
Teach colors, with the result displayed on the four green Output LEDs.
Section 4 – Exploring ColorSense
By clicking on the icons in the Sort Editor screen, you can:
• Upload Teach data from the sensor
• Download Teach data to the sensor
• Save Teach data to a file
• Read Teach from a file.
Also, you can use the Arrow buttons to
“move” Sort data up or down in order to
change the output assigned to a taught color.
The “X” button deletes the data from the
selected output.
The Analyze screen shows the predicted
performance for the Teach data, which is
shown as the number of standard deviations
separating each Sort Output. Values less
than 3 indicate that the sensor has a hard
time differentiating between the target
colors, and values greater than 5 indicate a
high degree of confidence.
By selecting Custom from the pull-down box, you
can manually adjust the weighting factors and see
the affect on performance. This unique capability
allows the RGB and Intensity weightings to be
optimized to the actual Teach data.
CAUTION: In order for the standard deviation prediction to be correct, it is essential that the Teaching be
done optimally. All eight Teach readings must be taken, the eight readings must include representative
samples of the range of colors that are considered permissible for that Sort Output, and the sample objects