Integrated Memory Circuit Having an Improved Logic Row Selection Gate

The invention relates to an integrated memory circuit which is divided into a number of memory blocks, each of which comprises memory cells which are arranged in rows and columns, the memory cells which are arranged in a column being connected to a column selection line while in several memory blocks the memory cells which are arranged in a row are connected to a row selection line, a row of memory cells in a memory block being activatable via a logic row selection gate whereto a row selection signal RS and a block selection signal BS are applied.

A memory circuit of this kind is known from ISSCC, Digest of Technical Papers, February 1983, pages 58-59. In this circuit the block selection signal BS and the row selection signal RS are applied to an AND-gate for the selection of a row of memory cells in a memory block. In practice an AND-gate consists of an inverting AND-gate whose output is connected to an inverting amplifier so that the circuit is large (6 transistors) and slow (due to 2 gate delays). A further possibility consists in the selection of a row in a memory block by means of an inverting OR-gate which should in that case receive the inverted block selection and row selection signals BS and RS. When CMOS transistors are used, however, this solution again results in a large and slow block selection circuit, because the PMOS transistors must be very wide in order to achieve the same power supply capacity (as the AND-gate), so that the input capacitances are high. The switching of the OR-gate is fast, but more time is required for charging these input capacitances. For example, in a 256 k memory the block selection signal BS has to drive 256 (or 512, 1024, depending on the memory organization) row selection gates in parallel.

An integrated memory circuit in accordance with the invention is characterized in that the logic row selection gate comprises two series-connected transistors, a first main electrode of a first transistor thereof receiving the block selection signal BS, a second main electrode thereof being connected to a row selection line as well as to a first main electrode of the second transistor whose second main electrode is connected to a reference potential point (ground) while its gate electrode receives the inverted block selection signal BS.

A preferred embodiment of an integrated static memory circuit including CMOS transistors in accordance with the invention is characterized in that there is provided a third transistor whose main electrodes are connected to those of the second transistor, the first transistor being of the P-MOS type, the second and the third transistor being of the N-MOS type, the gate electrode of the second transistor receiving the inverted block selection signal BS, the inverted row selection signal RS being applied to the gate electrode of the first and the third transistor. Such a memory circuit offers the advantage that, when another row is selected in the same memory block, it is no longer necessary to discharge the previously selected row selection line via the second transistor driven by the inverted block selection signal BS; this means that the memory is faster in accepting and supplying data. The use of CMOS transistors (PMOS and NMOS transistors) offers the known advantage that the memory circuit has a very low energy dissipation.

Embodiments of the invention will be described in detail hereinafter with reference to the accompanying drawing, in which:

FIG. 1 shows a memory circuit with block selection in accordance with the invention;

FIG. 2 shows a detail of a memory block as shown in FIG. 1; and

FIG. 3 shows a gate circuit for the memory block shown in FIG. 2.

FIG. 1 shows a memory circuit 10 whose memory section is divided into eight memory blocks 1 to 8. Each memory block comprises memory cells which are arranged in rows and columns and which are selected by means of a column address CA and a row address RA. The row address is applied to a row decoding circuit 11 and a first part of the column address CA is applied to the column address circuits 12a and 12b. The further part of the column address CA is applied to a block selection circuit 13, a memory cell in a memory block being selectable by means of the addresses CA and RA as will be described hereinafter. As is apparent from FIG. 1, a memory cell is selected not only by means of the row and column decoding circuits 11, 12a and 12b but also by means of the block selection circuit 13 which applies the block selection signals BS1, BS1 to BS8, BS8, respectively, to the respective memory blocks 1 to 8 for this purpose. The reason for such addressing in a memory circuit will be explained with reference to FIG. 2.

FIG. 2 shows a detail of the memory circuit shown in FIG. 1. The memory blocks 1, 2 to 8 are shown with the rows and columns of memory cells in each memory block. Row 1 of memory cells in block 1 is denoted by the reference 1C1. The first cell of this row is denoted by the reference 1C11, the second cell by reference 1C12, etc. The same code is used for the cells and rows in the further memory blocks. A memory cell in a block is selected as follows: via the column lines CA1 or CA2 or . . . CAn, one column is driven in each block. The lines CA1 in all the blocks can be driven by the same circuit, but can alternatively be driven by a separate buffer. The selection of a row of memory cells in a block is performed by means of the selection gate P which receives a row selection signal via the line XS1 or XS2 or . . . XSm for this purpose. The selection gate P in memory block j (1.ltoreq.j.ltoreq.8) and in row i of the memory block j is denoted by the reference Pji. Furthermore, the block selection signal BSj and the inverted block selection signal BSj are also applied to the gate circuit. Because the row selection signal XSi need not drive all memory cells of all rows iC1, iC2, . . . iCn (1.ltoreq.i.ltoreq.n), but only a number (8) of selection gates P, substantially faster driving of a row in a block is achieved. Because a large number of such logic selection gates P are present in the memory, that is to say one per row memory block, the logic selection gates P should not require an (excessively) large surface area. In accordance with the state of the art, for these gates use is made of AND-gates which comprise six transistors in the case of CMOS transistors. This gate circuit requires a large integration surface area. The NOR-gate in FIG. 2 requires a much smaller surface area and includes a substantially smaller number of transistors as is apparent from FIG. 3. The row selection gate Pji shown in FIG. 3 includes two series-connected transistors 31 and 32, a further transistor 33 being connected in parallel with the transistor 32. The transistors 32 and 33 are N-channel MOS transistors and the transistor 31 is a P-channel MOS transistor. The gate electrodes of the transistors 31 and 32 receive the row selection signal XSi and the gate electrode of the transistor 33 receives the block selection signal BSj. If transistor 31 is a N-MOS transistor, it should receive the non-inverted row selection signal XSi on its gate electrode. It will be apparent that the row selection line Lji which activates row i in a block j carries a high signal when the signal BSj is high and the signal XSi is low. The signal on the selection line Lji is low when either the signal BSj is high or the row selection signal XSi is high. In the embodiment of the gate selection circuit Pji is shown, the transistor 31 has a channel width of 8 micrometers, the transistor 32 a channel width of 4 micrometers, and the transistor 33 a channel width of 2 micrometers. It will be apparent that such a small gate circuit represents a low capacitive load for the control lines BSj, BSj and XSi, so that these gate circuits will cause only a small delay which is, of course, advantageous.

It is to be noted that the transistor 32 can be dispensed with if the block selection signal BSj is briefly made "low" after each selection of a memory cell and before a subsequent selection of a memory cell in another row within the same block, so that the row selection line Lji is discharged. Even though the selection gate Pji is thus slightly further reduced, the necessity of making the block selection signal BSj "low" (BSj is "high") reduces the speed of operation of the memory which is, of course, undesirable.

It is also to be noted that, even though the foregoing description concerns only a division in the direction of the rows (each row selection line is actually replaced by a number of row selection lines in the various blocks), a similar division could be made in the column direction. Identical selection gates could then be used.

The division of a memory into memory blocks with block selection (in the row and/or column direction) is notably attractive for 128 kbit or larger memories (256 k etc.). It will be clear that the terms "rows" and "columns" hereinabove are understood to be interchangeable.