The read-out step from a CCD is beautifully choreographed. The basic idea is illustrated in Figure 7. Charge accumulated in an MOS structure can be displaced sideways by using a synchronised sequence of gate voltages, as shown in the figure – rather like passing buckets of water down a line of poorly resourced firefighters.
Data are taken out of a CCD array in an orderly fashion, rippling through each column in turn. To do this requires two more electrodes per pixel, one on each side of the thin-gate photocapacitor, and an additional column of electrodes that forms an output register. These new electrodes are all parts of MOS structures that are ‘blind’. Their role is to shift sideways the charge stored in a line of pixels, so that all rows can be simultaneously displaced, say to the right, by one column. At the right-hand edge the output register column can be simultaneously loaded with data from all rows in a parallel action. Before overwriting with the input from the next column, the output line is marched downwards, one pixel at a time, passing serially through a single amplifier that converts stored charge to a proportional voltage signal that is subsequently digitised and sent on for storing in memory.
One of the performance advantages of a CCD is that data from every pixel are passed through the same amplifier. Two disadvantages are that transfer into memory is a serial process, and that during read-out it is necessary to freeze the image frame to avoid overwriting during the read-out cycle. This is a major factor influencing how quickly successive photos can be taken.