Department of Computer Science Course: CS 3724

Read only memory (ROM):

Often, today, devices requiring 8 or more input variables are implemented using a ROM. A simple type of ROM can be constructed from a decoder, a MUX, and a number of wires. We will look at a small (16 bit) ROM constructed in this way. Normally, memory is arranged in a square array, as shown in Figure 1.19. To use the ROM to implement a logic function, the address lines are used as the variable inputs, and the contents of the memory are the function values. Usually the memory has a word length of more than 1 bit; typically 4 or 8 bits, so several functions can be implemented simultaneously.

In Figure 1.19, it is assumed that the decoder produces a logic 1 as output when the input code selects that output; otherwise it produces logic 0, and that a logic 1 output ``outvotes'' a logic 0 output, in the sense that if both are present on the same wire, the logic 1 will dominate. (``Real'' circuits usually have the opposite behavior). A bit is ``programmed'' when a link is present.

Figure 1.19: A read-only memory implementation

In the example of Figure 1.19, the function is

$$Y = \overline{A} \cdot \overline{B} \cdot {C} \cdot \overlin... ...cdot {D} + {A} \cdot \overline{B} \cdot {C} \cdot \overline{D}$$

corresponding to memory locations 0010, 0110, 1001, and 1010.

This general organization is used for other types of memory, as well. One type of programmable read-only memory uses small ``fuse links'' to connect the horizontal and vertical wires at each intersection. This device is ``programmed'' by passing sufficient current through a link to ``blow'' the fuse. The link could also be a transistor which could be turned on or off, allowing a read-write type of memory to be implemented. We will see logic devices which could be used for this purpose shortly.