Page 458 - DSP Integrated Circuits
P. 458

10.4 Storage Elements                                                443


        latch (with J = S and K = R) except that the response to J = K = 1 is that the latch
        changes state. This is expressed as



            The next state of a D latch is the same as the current input. A JK flip-flop can
        be changed into a D flip-flop by letting D = J = K.
            The edge-triggered flip-flop shown in Figure 10.12 may only change state
        when the clock signal changes. Changes in the input signal during the clock cycle
        have no effect on the stored state, which depends solely on conditions prevailing
        immediately before the active clock edge.
            A master—slave flip-flop can be
        implemented by cascading two latches
        controlled by two different clock signals.
        Both the type of latch and the relation-
        ship between clock signals vary in differ-
        ent implementations. In its simplest
        form, two latches are driven by a single
        clock signal, (£3 being derived from <&i by
        passing it through an inverter. A com-   Figure 10.12 Edge-triggered flip-flop.
        mon variation is to use two independent
        clock signals,  <£>i and  d>2, that are
        nonoverlapping.
            Inputs to an edge-triggered flip-flop must be stable before the arrival of the
        clock edge. The required time interval is called the setup time for the inputs,
        which includes a delay due to possible clock skew.



        10.4.2 Dynamic Storage Elements
        The input capacitance to a gate built of MOS transistors represents a significant
        capacitance that can be exploited as a dynamic memory element. Dynamic logic
        circuits rely on temporary storage of charges in gate capacitances to simplify and
        speed up the circuits. However, the state (charged or uncharged node) of a
        dynamic node is retained only for a short time (ms), because of leakage currents.
        The node must therefore be recharged periodically.
            The basic principle of dynamic storage
        is illustrated in Figure 10.13. A control sig-
        nal, called a clock signal, controls a switch.
        When the switch is closed, input signal Vj n
        is applied to the input of a logic circuit, in
        this case an inverter. During this phase,
        neglecting any delays in the switch and the
        inverter, output signal V out is equal to the
        inverse of Vi n. After the switch has been
        opened by the control signal, the output sig-
        nal will remain at the same value, since the
        input to the inverter will remain the same  Figure 10.13 Principle of a dynamic
        due to the charge stored in the gate capaci-          storage element
        tance, C gate.
   453   454   455   456   457   458   459   460   461   462   463