Manufacturers of CCD sensors grade devices according to the number and type of defective pixels. The manufacturing yield of each sensor grade strongly affects the CCD cost; the more perfect the sensor, the higher its cost. Because the CCD sensor is a large-cost component in the overall camera system, the choice of sensor grade is an important consideration when purchasing a camera. Unfortunately, each CCD manufacturer uses a different scheme to grade devices. Grading schemes typically run from a grade 0 device, designating the highest quality available (nominally defect-free), to grade 1, 2, or 3, with the number of defects increasing with the grade number. To assist customers in choosing the proper grade, we've provided the definitions below, as well as a few examples to compare grading between manufacturers.
The central zone is an area in the middle of the CCD array. The exact location and size varies with the manufacturer. Defects in this region are usually specified separately from the overall number of defects.
This is the group of pixels surrounding the defect in question, usually 10,000 pixels or less. Again, the exact specification is manufacturer-specific.
A point defect is a pixel whose response differs by ±N% compared to the mean values of all pixels in the neighborhood. "N" can be as low as 6% or as high as 20% depending on the manufacturer.
This is a group of adjacent point defects. The maximum allowable number of defective pixels in a cluster varies between 3 and 9, depending on the manufacturer.
Column or Row Defect:
A column or row defect refers to a column or row, or partial column or row, whose response varies by at least ±N% from the neighborhood mean value. "N" is usually the same number as for point defects.
A trap is a pixel that traps charge during the charge-transfer process. Charge transfers out of the trap at a lower rate, leading to charge being "left behind." Once a trap is filled, a steady state is reached where it no longer consumes signal electrons. Some manufacturers give specifications for both the number of low-level traps (filled with typically <2000 e-) and high-level traps (filled with typically <10,000 e-). The physical location of the trap is also important, particularly for low-light applications. Traps in the serial register of the CCD can affect signal from nearly the entire sensor. Traps in a column only affect that column's signal. Traps are often quite dependent on the CCD's operating temperature.
Some defects (pixel, cluster, or column) are substantially brighter than adjacent regions. Often, this is due to higher-than-average. These defects tend to disappear as the device is cooled. Because their location and dark current rate are constant, they can often be compensated for by dark current subtraction.
CCD Defect Specifications
for Several Selected CCDs and Manufacturers
|Grade 1 Specifications (maximum number in total)|
|Point Defects||10 hot pixels
3<3 dark pixel cluster
2<5 dark pixel cluster
|Column Defects||1 dark,
|Grade 2 Specifications (maximum number in total)|
|Total Defects||15 hot pixels
15<3 dark pixel cluster
8<5 dark pixel cluster
1<10 dark pixel cluster
|Column Defects||6 black,
This table compares some of the defect specifications for a few popular CCD arrays. The defect amounts are defined by the manufacturer in terms of number of pixels, columns, or clusters whose response differs by ±N%. The deviation, N, is defined independently by each CCD manufacturer and definitions vary widely. Contact Roper Scientific for the exact definitions of the device you are considering.