Switching Matrix

a noncontact discrete-action switching device with n inputs and m outputs that are connected in such a way that definite combinations of signals at the inputs uniquely correspond to definite combinations of signals at the outputs. It is used chiefly in digital computers as an encoder, in which a signal at one of the inputs excites several outputs simultaneously, and as a decoder, in which a definite combination of signals at the inputs excites just one output. The device received the name “matrix” from the way it is represented as a network of intersecting horizontal and vertical buses—a network of rows and columns. This representation sometimes corresponds to the device’s external appearance.

Simple switching matrices can be constructed from resistors, transformers, and capacitors that connect vertical and horizontal conductors at selected points of intersection. Resistive, inductive, and capacitive couplings are linear. It is therefore assumed that the input signals are discrete (binary) in nature and the m output reading devices have a sharp switching threshold. Switching matrices of this type are used extensively in memory units for storing subroutines, constants, and arithmetic and other tables.

Nonlinear elements are often used in switching matrices: semiconductor diodes, diode matrices, transistors, and magnetic cores with rectangular hysteresis loops. In this case the basis of the switching matrix is AND circuits (or inhibit circuits) and sometimes a combination of logical circuits, which perform, respectively, the logical functions of conjunction and disjunction. In computer technology such switching matrices are used in code translators—for example, to translate telegraph code into computer code and vice versa. They are also used in combination shifters, adders, and multipliers. Switching matrices based on magnetic cores are used in memory devices for address selection.

Figure 1. Switching matrices: (A) with diodes, (B) with ferrite cores; (R) resistors, (E₁₁) power supply, (D) diodes, (F) ferrite cores (toroids), (I_r) read current, (I_c) write current, (a), (b), and (c) input variables

Figure 1, A illustrates a diode switching matrix for the summing of three binary signals. A sum signal appears on one of the four lower buses only if one or all of the input variables are equal to 1. A signal appears on the carry buses when two or three variables are equal to 1. A similar adder based on toroidal magnetic cores with a rectangular hysteresis loop is shown in Figure 1,B. The horizontal lines represent cores, and the vertical lines windings. The diagonal slashes show which input variables’ windings relate to the given core. Let us assume the cores are first all magnetized in one direction. When a reading signal is fed concurrently with input signals representing binary variables, the core whose windings have no inhibit current will undergo a reversal of magnetization.

The most important parameters of a switching matrix are speed (switching speed) and the ratio of amplitude of the useful signal to the amplitude of the noise. The speed varies from microseconds to nanoseconds depending on the type of elements used. The signal-to-noise ratio usually ranges from 10 to 20.