Variations between Watec WAT-120N cameras Version 21. January 2004
Martin Elsässer

For a short while i have several Watec WAT120N cameras available. I have done some comparisons, to see how the properties of the cameras vary among the same model.

Dark-images compared

I have taken dark images with all three cameras under favorable conditions (outside temperature 1°C). The settings of the used framegrabber where identical for all cameras. The serial numbers of the three cameras were 23,51 and 70.
The 51-camera had a small foam-pad removed, which usually sticks the backside of the CCD-chip to the circuit board. This did not seem to have any significant effect, though.

Histograms of the dark-images of the three cameras (compressed in y-direction by sqrt() function). The histogram plots a frequency of occurrence of a certain brightness value against the brightness value. Each dark-image is the median of 35 individual images for each camera.

The histograms show, that there is a certain amount of variation between the three cameras, with 70 and 51 being quite similar, despite the small modification to the 51 camera.
Standard deviation of the pixels from the dark-images. For each pixel, the median brightness was established over the range of 35 images. All pixels with the same median-value were combined, to find the standard deviation. Generally, the lower the value, the less noise the system has.

Apparently, the later the camera was produced, the higher the noise? Strange.

Noise across the chip

For each camera, i mapped the variations of the individuals pixel brightness during the 35 dark images taken, to create some kind of map of the distribution of the noisy pixels. The resulting pictures were quite similar for all three cameras, with regular vertical bars of increased noise.

Standard deviation of the pixel brightness for the 23, with some comsics creating bright dots.
Standard deviation of the pixel brightness for the 51, with some comsics creating bright dots.
Standard deviation of the pixel brightness for the 70, with some comsics creating bright dots.

I used a rather simple USB 2.0 framegrabber for these tests, which i suspect to have some influence on the signal quality. This might greatly influence the actual noise values seen above.