There are two versions of the IS32 OpticRAM: the IS32 and IS32A. Beginning in September 1983, the IS32 was replaced in favor of the IIS32A. The only difference between the two devices is size. The IS32A is exactly 20 percent smaller in the horizontal and vertical dimensions. The dimensions below are for the IS32A. To calculate dimensions for the larger IS32 device, multiply by 1.25.
Broadband sensitivity of the IS32 OpticRAM is approximately 2uJ/sa cm.
Silicon detectors have a useful optical sensitivity over the region of the spectrum is which silicon absorbs photons. This extends from 200 nanometers to 1100 nanometers. However, a complete characterization of the IS32 is still under vIZy. The sensitivity follows the silicon characteristic curve since the IS32 is built using silicon. The IS32 is impervious to damage by high light intensity. It has a high quantum efficiency and a binary output that is proportional to the amount of incident light and integration time (referenced to a threshold). However, oversaturation of the IS32 by more than 4 F-stops will, for the duration of oversaturation, make the first half of the array all light and the other half all dark. This is only a temporary situation for the duration of the saturation. The IS32 is sensitive up to near UV.
The IS32 chip is mounted in the package with 20 mils tolerance in both the X and Y axis. This suggests that if an OpticRAM package is replaced in a camera, a physical realignment of the camera to the scene is necessary. The tolerance from surface of the array to the lens mount from camera to camera is 20 mils with a 6 degree rotational tolerance.
If you access a cell (pixel) in the OpticRAM using an address of zero for both the row and column, the OpticRAM will not physically select Row 0 and Column 0. This is because the internal address decoding does not provide a one-to-one correspondence between the address count and the physical row and column. A simple circuit, consisting of a 7486 and a 7404, performs the necessary code conversion to achieve the desired one-to-one correspondence. See Figure D-3
One of the primary goals in designing a low cost integrated circuit such as the OpticRAM, is to minimize its physical size. To achieve this goal, the cells in the OpticRAM are arranged in an interleaved pattern. If an image is read out of the OpticRAM by counting successively down the rows and columns, the image will look "fuzzy" around the edges because the pixels will be slightly misplaced in the graphics matrix.
To accomodate the pixel misplacement, the data from the optics must be mapped into the graphics matrix so that the arrangement of the pixels in the graphic matrix matches the physical arrangement of the cells in the OpticRAM. Due to the interleaved cell pattern on the OpticRAM, the array is much longer than it is wide, resulting is spaces between the cells in the column direction. Because of the spaces, the 128 X 256 array of cells will map very nicely into a 129 X 514 matrix. We will call this matrix the Cell Placement Grid.
The cell placement grid is shown in Figure D-2 below. For a single array, there are a total of 129 rows and 514 columns. Only the corners of the array are shown. The placement grid indicates where the information from each cell in the OpticRAM should be mapped. For instance, if the cell at address Row 1, Column 1, in the OpticRAM is read, the value (a 1 or 0) should be placed in the placement grid at location X=2, Y=3.
When every cell has been read and the values placed in the appropriate locations, about half of the grid remains empty. We will call these empty locations "space pixels." The space pixels can be set all high or all low to provide a light or dark background for the image. Another alternative is to set each space pixel to the level that agrees with the majority of its nearest neighbors. For example, let's say the pixel at grid locations X=2,Y=2 (R1 C1) and X=3,Y=1 (R1 C0) are high, and the pixel at grid location X=3,Y=3 (R3 C0) is low. These are the three nearest neighbors of grid location X=3,Y=2. The majority of these nearest neighbors is high, so the previously empty grid location X=3,Y=2 is set high also. This technique can be applied to all empty grid locations except those near the edge of the array. A modified technique can be used for these edge space pixels, although there is less optical data to work with. Another alternative is to simply not use the edge rows and columns.
Having the cells laid out in the IS32 the way they are, gives the IS32 much greater resolving power than if the cells were laid out linearly.