Background of the Invention
This invention relates to a Random Access Memory (RAM) of the static type.
A memory cell of a static RAM is, in general, mainly composed of a flip-flop, and the higher the memory density becomes, the smaller the cell area is. Therefore, the conductance (gm) of the driver transistor in each cell is lowered, and, as a result, the bit line drive capability is reduced, requiring a small conductance (gm) load transistor for each bit line. However, such a static RAM has a disadvantage in that when the conductance (gm) of the bit line load transistor is small, the charging capability of the bit line becomes insufficient, and therefore the reverse data readout speed becomes low in a readout cycle immediately after a write cycle (i.e., where a "1" must be read immediately after a "0" has been written via the same bit line pair, or when a "0" is read immediately after a "1" has been written).
Summary of the Invention
An object of the present invention is to avoid this disadvantage by providing one bit-line-charging transistor in parallel with each of the bit line load transistors. The bit-line-charging transistors are temporarily turned ON after the write operation, thus connecting both bit lines of each bit line pair to the power supply voltage, in order to promptly recharge them.
Brief Description of the Drawing
The present invention will be explained below with reference to the accompanying drawings, wherein:
FIG. 1 is a circuit diagram schematically illustrating one column of static memory cells MC.sub.1, MC.sub.2, etc., connected to a pair of bit lines DB and DB, to which bit-line-charging transistors Q.sub.1 and Q.sub.4 have been connected in parallel to the bit line load transistors Q.sub.2 and Q.sub.3, respectively, in accordance with the present invention;
FIG. 2 is a graph illustrating operation signal waveforms, which appear at various points in the diagram of FIG. 1 during a reverse data read out operation; and
FIG. 3 schematically illustrates the structure of the control signal generator, which is used to generate the signal WE' for controlling the bit-line-charging transistors in accordance with the present invention.
Description of the Preferred Embodiment
FIG. 1 shows an embodiment of the present invention. Elements other than the transistors Q.sub.1 and Q.sub.4 are conventional elements. The transistors Q.sub.1 and Q.sub.4 are used for charging bit lines and are connected in parallel with the load transistors Q.sub.2 and Q.sub.3 of the bit lines DB and DB, respectively. In the preferred embodiment, the transistors Q.sub.1 and Q.sub.4 are MOSFETs. The bit lines DB and DB are connected to a plurality of static type memory cells MC.sub.1, MC.sub.2 . . . , which are selected by the word lines Xn, Xn+1, . . . .
Yn, Yn+1, . . . are signals for bit line selection and when the transistors Q.sub.13, Q.sub.14 turn ON with the signal Yn, the bit lines DB and DB are selected. The transistors Q.sub.5, Q.sub.8, Q.sub.9 and Q.sub.12 of cells MC.sub.1, MC.sub.2 are the transfer gates, which are turned ON and OFF by the voltage level of the word lines Xn or Xn+1. The cells MC.sub.1 and MC.sub.2 are essentially composed of a flip-flop circuit including load resistances, R.sub.1, R.sub.2, R.sub.3 and R.sub.4, and driver transistors Q.sub.6 and Q.sub.7 and Q.sub.10 and Q.sub.11. The buffer G.sub.1 and inverter G.sub.2 form a write circuit. When the buffer G.sub.1 is enabled by the write enable signal WE while the data input D.sub.IN is H (high) level the potentials of the bit lines become as follows: DB=L (low), DB=H (high). At this time, if the gates Q.sub.5 and Q.sub.8 of the cell MC.sub.1 have been turned ON, the transistor Q.sub.6 turns ON while Q.sub.7 turns OFF, allowing write operation to the cell MC.sub.1.
A reverse data read out operation will now be described, to indicate the operation of the present invention, particularly as compared with other devices. FIG. 2A shows word selecting signals X.sub.n and X.sub.n+1, which select memory cells MC.sub.1 or MC.sub.2, respectively, within the column shown in FIG. 1. (Of course, word address buffers, which are not shown, would routinely be used to provide the signals X.sub.n and X.sub.n+1.) The specific example which will now be discussed assumes that a "0" is written into cell MC.sub.1, and immediately thereafter a "1" is read out from memory cell MC.sub.2. The signals associated with these two successive operations are shown in FIG. 2 beneath the headings write cycle .circle.1 and read cycle .circle.2 , respectively. Thus, in write cycle .circle.1 the bit lines are respectively set at DB=L (low), DB=H (high), and in read cycle .circle.2 the bit lines are respectively set at DB= H, DB=L.
At the time of write cycle .circle.1 , the write enable signal WE becomes L, making DB=L, DB=H in accordance with the data input D.sub.IN. In this case, the level L is usually lowered to almost O V in order to maximize the write amplitude. Therefore, this L level of bit line DB at this time is very much lower than the L level (about 2.5 V) at the time of the readout operation, and in a reverse data readout operation, a large voltage swing must accordingly be forced on bit line DB.
In the conventional circuit structure, the bit line DB is charged only by the load transistor Q.sub.2 from the power supply V.sub.cc at the time of read cycle .circle.2 , but since the conductance (gm) of Q.sub.2 is low due to the high integration density, the potential of DB merely increases gradually, as indicated by the dotted line in FIG. 2D. Thus, the time t.sub.3 where the potential levels of the bit lines DB and DB cross and, as a result, the timing t.sub.4 where the potential levels of the data outputs D and D cross each other is delayed, thereby slowing the readout operation.
In order to solve this problem and improve the readout operation speed, the present invention provides the transistors Q.sub.1 and Q.sub.4, connected in parallel with the load transistors Q.sub.2 and Q.sub.3, which are temporarily turned ON when the write operation ends (i.e., when the write enable signal WE rises), so that the bit lines DB, DB are charged from the power supply V.sub.cc via the respective pairs of parallel transistors Q.sub.1 and Q.sub.2, and Q.sub.3 and Q.sub.4. This has the same results as if the bit line were charged by a load transistor having a large conductance (gm), and it becomes possible to quickly raise the bit line potential.
The bit-line-charging transistors Q.sub.1, Q.sub.4 are controlled by the signal WE' (FIG. 2C) which is generated by the circuit shown in FIG. 3. G.sub.3 is an inverter having an output connected to one input of the NOR gate G.sub.5. Simultaneously, the output of the inverter G.sub.3 is also connected to the input of a delay type inverter, comprising an inverter G.sub.4 and a capacitor C.sub.1. The output of this delay type inverter is fed to the other input of the NOR gate G.sub.5. The signal WE' obtained from NOR gate G.sub.5 is opposite in phase to the signal WE, and has a waveform which remains at the H level only for a certain period, determined by the capacitance of capacitor C.sub.1, after the falling edge of the signal WE. Thus, because the transistors Q.sub.1 and Q.sub.4 are turned on by the signal WE', the bit line DB having the L level is rapidly charged after the end of the writing operation, and the potential increases as indicated by the curve A in FIG. 2D. The degree of increase of the potential depends on the width of the signal WE' and therefore the width of WE' is selected such that the bit line DB is increased from the write operation L level to the readout operation L level. In other words, when the write operation ends, the level L of the bit line DB is immediately set to the level L for the read operation. Thus, when the write cycle .circle.1 ends, the bit line DB increases in level from the L level of the write cycle .circle.1 and the reverse data readout (in the readout cycle .circle.2 starts with DB at the readout level L. Therefore, the potential levels of the bit lines DB, DB cross at the time t.sub.1 (preceding the time t.sub.3), while the potential levels of the data outputs D, D cross at the time t.sub.2 (preceding the time t.sub.4), so that the readout operation speed of the memory cells is improved.
As explained above, it is desirable that the period for keeping the bit-line-charging transistors Q.sub.1, Q.sub.4 ON be only temporary, at the end of a write cycle .circle.1 . If the transistors Q.sub.1, Q.sub.4 are kept ON during the read cycle .circle.2 , a large current would flow into the memory cells from the power supply V.sub.cc, causing increased power consumption. Moreover, since the gate voltage of the transistor Q.sub.10 (which is OFF during read cycle .circle.2 ) in the memory cell MC.sub.2 of FIG. 1, is raised by the power supply V.sub.cc if the transistor Q.sub.4 is kept ON for a long period of time, the flip-flop comprising the transistors Q.sub.10 and Q.sub.11 is liable to be inverted by the influence of electronic noise, whereby information stored in cell MC.sub.2 may also be inverted during the read cycle .circle.2 . These disadvantages can be eliminated by turning ON the transistors Q.sub.1 and Q.sub.4 only temporarily, by applying the pulse WE' to their gates after a write cycle.
As explained previously, the present invention offers an advantage that a high readout operation rate can be maintained even when the cell area is reduced in the static RAM.
The features and advantages of the invention are apparent from the detailed specification, and thus it is intended by the appended claims to cover all such features and advantages of the memory which falls within the true spirit and scope of the invention. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation shown and described, and accordingly all suitable modifications and equivalents may be resorted to, falling within the scope of the invention.