Semiconductor Dynamic Memory

The present invention relates to a semiconductor dynamic memory which is capable of reducing a cycle time.

The dynamic memory essentially requires a reset period. In the conventional dynamic memory in which all of the portions are reset at one time, the cycle time is the sum of a period (access time) including from the start of access of a portion to which the access is made at first to the completion of access of a portion to which the access is finally made and a reset time. On the other hand, the static memory does not require resetting. Therefore, the cycle time of the static memory is nearly equal to the access time. Thus, the dynamic memory having a long cycle time is capable of writing and reading data in a quantity smaller than that of the static memory within a unit period of time.

An object of the present invention is to provide a semiconductor dynamic memory which removes the drawbacks in the conventional dynamic memory.

Another object of the present invention is to provide a semiconductor dynamic memory which is capable of reducing a cycle time.

A further object of the present invention is to provide a semiconductor dynamic memory having a cycle time which is equal to, or shorter than, an access time.

For achieving the above mentioned object, the present invention deals with a semiconductor dynamic memory comprising a plurality of functional blocks interface circuits such as a row-enable buffer, a row address buffer which receives the output signal from the row enable buffer, a word decoder which is connected to the row address buffer, a group of sense amplifiers which are coupled to word lines connected to the word decoder, a column enable buffer, a column address buffer which receives the output signal from the column enable buffer, a column decoder which receives the column address signal from the column address buffer and which selects one of the sense amplifiers, a data buffer which receives the output of the selected sense amplifier, and an output buffer which is connected to the data buffer. At least one of the functional blocks is reset to be ready to execute the next processing by a signal which is provided from a subsequent functional block and proves that the subsequent function block has begun its operation.

Further features and advantages of the present invention will be apparent from the ensuing description with reference to the accompanying drawings to which, however, the scope of the invention is in no way limited.

FIGS. 1 and 2 are a block diagram and a time chart illustrating the construction and operation of a major portion of a conventional dynamic memory;

FIGS. 3 and 4 are a block diagram and a time chart illustrating an embodiment and operation according to the present invention;

FIGS. 5 and 6 are a diagrams illustrating in detail the row-enable buffer circuit of FIG. 3 and waveforms for illustrating its operation;

FIG. 7 is a diagram illustrating in detail the word decoder, the sense amplifiers, the column decoder and the writing system circuit shown in FIG. 3;

FIGS. 8A, 8B and 8C are diagrams illustrating in detail the column decoder, the buffer data and the output buffer shown in FIG. 3; and

FIGS. 9A, 9B and 9C are diagrams illustrating waveforms for illustrating the operation of the circuit shown in FIGS. 8A, 8B and 8C.

FIGS. 1 and 2 illustrate the construction and operation of a major portion (peripheral circuit) of the interface circuit of the conventional most generally employed dynamic memory. As an inverted signal RAS of a row-address strobe assumes the L (low) level, the circuit in the row system commences to operate, and the row-enable buffer (REB)1, row-address buffer (RAB)2 and word decoder (WD)3 produce outputs RE, RA and WL, successively. Then, as an inverted signal CAS of the column-address strobe assumes the L level, the column system commences to operate, whereby the column-enable buffer (CEB)4, column-address buffer (CAB)5 and column decoder (CD)6 produce outputs CE, CA and D, successively. Here, one of the outputs BD of the sense amplifiers 7 produced by the operation of the row system is selected by the column decoder 6, and is converted into a data out DO via data buffer (DB)8 and output buffer (OB)9. At the time when the operation of the output buffer (OB)9 is finished, the inverted signals RAS and CAS assume the H (high) level. Therefore, the row-enable buffer (REB)1 and column-enable buffer (CEB)4 produce reset signals RE and CE, so that the row enable buffer (RAB), the word decoder (WD), . . . the column address buffer (CAB) and the column decoder (CD) . . . are reset at one time. In FIG. 2, time axis numerals 0, 50, 100, . . . at the upper most line represent the lapse of time in unit of nanoseconds. According to this embodiment, therefore, the cycle time is 270 nanoseconds. On the other hand, the time t.sub.RAC from the start of access to a moment at which the read data RD is produced is 150 nanoseconds. Therefore, the cycle time is considerably longer than the time t.sub.RAC.

FIGS. 3 and 4 illustrate the construction and operation of the interface circuits of the a major portion of memory according to the present invention. As illustrated in the diagrams, the individual portions according to the present invention are reset immediately after the operation is finished, and are ready to start the next operation. That is, the functional blocks are reset as soon as the subsequent block has begun operation and captured the signal. Namely, the row system and column system commence the operation when the inverted signals RAS and CAS assume the L level. Here, however, the row-enable buffer (REB)11 is immediately reset by a signal which is caused by the operation of the row-address buffer (RAB)12. The same also holds true for the row-address buffer (RAB)12 and column-enable buffer (CEB)14, without the need of waiting for the return of signals RAS and CAS. Therefore, when the reset is finished, the individual portions enter again into the active period and perform the next operation. Consequently, the cycle time becomes equal to the sum of the active period and the reset time, and is greatly reduced. Although it is difficult to perform the read-modify-write operation, the cycle time is so shortened that there practically arises no problem. Further, the dynamic memory which performs the address multiplex operation, must latch a row address as well as a column address, and hence necessitates two clock signals RAS and CAS. If a column address is latched by utilizing the rise in the clock signal RAS, however, the clock signal CAS need not be employed.

When the writing operation is taken into consideration, the word decoder (WD)13 must be reset after the completion of the operation of the column decoder (CD)16 which is a block of the next but one stage (the second stage--the first one is skipped). Further, the output buffer (OB)19 which generates read data to the output terminal D.sub.out receives a reset signal from the column decoder 16 when the column decoder 16 commences operation, starts the resetting operation, and completes the resetting operation while the data buffer (DB)18 is being operated. Thus, with the output buffer 19 being reset, it is possible to retain the read data of the previous cycle up to a moment just before a new read data is produced. In this case, the read data is maintained at the output terminal at all times, and therefore it is not allowed to commonly utilize the output terminal for another memory or to connect the output terminal in parallel with another memory. The output terminal can be connected in parallel with another memory if a chip select circuit (CSC)21 is provided and the output buffer 19 is controlled by the output signal CS. In FIG. 3, reference numeral 20 (WSC) denotes a circuit of the writing system, a signal WE denotes an inverted write-enable signal, and a signal D.sub.IN denotes a writing data. The time axis numerals 0, 50, 100 . . . shown at the upper most line of FIG. 4 denote the lapse of time (in nano-second units). The time t.sub.RAC from the first access to a moment at which the read data is produced is 150 nanoseconds, the same as the conventional memory shown in FIG. 2. On the other hand, the row enable buffer (REB)11 commences the reset at the time 40 nanoseconds and said reset is completed at the time 100 nanoseconds, so that at the time 100 NS the operation of the next cycle can be carried out. Therefore, one cycle can be completed during 100 nanoseconds. Further, the functional blocks other than the row enable buffer (REB)11 can complete similarly one cycle during 100 nanoseconds. Therefore, the memory can be operated with the cycle time of 100 nanoseconds. This cycle time of 100 nanoseconds is more rapid than that of the conventional memory which requires the cycle time of 270 nanoseconds, so that the function of the memory in the present invention is greatly improved.

FIGS. 5 and 6 illustrate a circuit and timings with regard to the row-enable buffer (REB)11. Symbols Q.sub.1 to Q.sub.14 denote MOS transistors or MOS capacitors, and N.sub.1 to N.sub.5 denote nodes or potentials at the nodes. When the inverted signal RAS assumes the low level, the node N.sub.2 assumes the high level, the transistors Q.sub.7 and Q.sub.8 are rendered conductive, the node N.sub.4 assumes the high level, the node N.sub.3 assumes the low level, the transistors Q.sub.10 and Q.sub.13 are rendered conductive, the transistors Q.sub.12 and Q.sub.14 are rendered non-conductive, and the node N.sub.5 and output RE assume the high level. When the output RE assumes the high level, the row-address buffer (RAB)12 operates to produce the output signal RA of the high level. The signal RA is fed back to the row-enable buffer (REB)11 in FIG. 5, whereby the transistors Q.sub.5, Q.sub.6 and Q.sub.9 are rendered conductive, the node N.sub.2 assumes the low level, the transistors Q.sub.7 and Q.sub. 8 are rendered non-conductive, the node N.sub.3 assumes the high level, the node N.sub.4 assumes the low level, the transistors Q.sub.10 and Q.sub.13 are rendered non-conductive, the transistors Q.sub.12 and Q.sub.14 are rendered conductive, and the node N.sub.5 and the output RE assume the low level. Here, since the signal RA is reset by the completion of the operation of word decoder (WD)13, the inverted signal RAS must be assumed to be high level before the signal RA is reset. If the inverted signal RAS is maintained at the low level, the row-enable buffer (REB)11 commences the next operation when the signal RA is reset.

FIG. 7 illustrates a circuit including the word decoder 13, the column decoder 16, the sense amplifier 17, the data buffer 18 and the writing system circuit 20 in FIG. 3. As seen from FIG. 3, almost all functional blocks receive the reset signal from the next functional block, however, the word decoder (WD)13 should receive the reset signal after the time when the function of the column decoder is completed. Next we will explain the reason why the word decoder (WD)13 should receive the reset signal from the column decoder which is a block of the next but one stage. As illustrated in FIG. 7, the sense amplifier 17 in FIG. 3 is formed by a group of sense amplifiers 17a, . . . , 17n, the column decoder 16 in FIG. 3 is formed by a group of column decoders 16a, . . . , 16n and the write system circuit 20 includes a writing circuit 20a and a buffer amplifier which includes transistors Q.sub.21, Q.sub.22, Q.sub.23 and Q.sub.24. In the circuit shown in FIG. 7, the output WL1, . . . , WLm of the word decoder are coupled via memory cells MC and bit lines BL1, . . . , BLn to the sense amplifiers 17a, . . . , 17n. The column decoder 16a, . . . , 16n receives the output (BD.sub.1, BD.sub.1, . . . , BD.sub.n, BD.sub.n) from the sense amplifiers 17a, . . . , 17n and the output signal CA from the column address buffer 15, and the output signal of the column decoder 16a, . . . , 16n are coupled via data lines DL and DL to the data buffer 18. In the reading operation, the word decoder (WD)13 selects one of the word lines WL.sub.1 .about.WL.sub.2m and the data of the memory cells which are connected to the selected word line are transmitted to the bit lines and amplified by the sense amplifiers SA.sub.1 .about.SA.sub.n, and only the data which is selected by the column decoder CD.sub.1 .about.CD.sub.n is transmitted to the lines DL and DL. When the reading operation is carried out in such a manner, the word decoder (WD)13 can be reset after the data of the memory cells are transmitted to the bit lines and amplified by the sense amplifiers. Therefore, it is not necessary to await the operation of the column decoder (CD)16 when executing the reading operation. Contrariwise, in the writing operation, the data which is written in the lines DL and DL by the writing system circuit 20 is written via the column decoder sense amplifiers and the bit lines in the memory cell which is selected by the word lines. Therefore, the word lines can not be reset till the time when the column decoder is operated and the data is written in the memory cells via the lines DL and DL and the bit lines. This is the reason why the word decoder (WD)13 receives the reset signal from the column decoder (CD)16.

As shown in FIG. 3, different from the other blocks or interface circuits, the output buffer (OB)19 receives the reset signal from the column decoder 16 which is two stages from the output buffer 19. This is due to the reason that the read data is maintained at the output terminal D.sub.out till the next data is output at said output terminal D.sub.out. If it is not necessary to output the data before the next data is output, the chip select circuit (CSC)21 may control the output buffer (OB)19 so as to disable the output D.sub.out. For the purpose of explaining the method for resetting the output buffer (OB)19, a more detailed functional block diagram of the column decoder, the data buffer and the output buffer are shown in FIGS. 8A.about.8C.

The column decoder 16 shown in FIG. 3 includes a column decoder driver 16a and a column decoder 16b as shown in FIG. 8A, the data buffer 18 shown in FIG. 3 includes a data buffer driver 18a and a data buffer 18b as shown in FIG. 8A, and the output buffer 19 shown in FIG. 3 includes an output buffer driver 19a and an output buffer 19b. The reset signal is supplied from the column decoder driver 16a to the word decoder 13, the sense amplifier 17 and the writing system circuit, the data buffer driver 18a generates a reset signal for the column decoder driver 16a and the column decoder 16b. The reset of the output buffer 19 is commenced by the output signal CDD of the column decoder driver 16a. On the other hand, the data buffer 18 commences operation by the signal CDD, so that the output buffer driver 18b is reset during the operation of the data buffer 18b. When the output buffer driver 18b is reset, the output OBD of the output buffer driver 18b is also reset so that the output buffer 19b is reset at the same time. The output buffer driver 19a and the output buffer 19b are competely reset till the time when the data buffer driver 18b outputs the output signal DBD. Therefore, the output buffer 19b can receive the signal DBD and the read data RD and RD, so that the output buffer 19 commences operation and supplies the read data at the output terminal D.sub.out. At the same time when the operation of the output buffer driver 19a is completed, the signal DBR is generated so as to reset the data buffer 18. This signal DBR is automatically reset by the timing circuit included in the output buffer driver 19a at about the time when the data buffer 18 is completely reset by the signal DBR. On the other hand, the output signal OBD of the output buffer driver 19a is maintained till the time when the output buffer driver receives the signal CDD in the next cycle so that the read data is maintained at the output terminal D.sub.out.

FIG. 8B is a circuit of the output buffer driver 19a, symbols Q.sub.31 through Q.sub.52 denote MOS transistors or MOS capacitors and symbols N.sub.11 through N.sub.19 denote nodes or potentials at the nodes. The circuit which is formed by the transistors Q.sub.31 .about.Q.sub.42 is the circuit which forms the signal OBD. The timing chart of this circuit is shown in FIG. 9A. At first, when the signal CDD is placed in the high level, the potential in the nodes N.sub.12 and N.sub.13 is placed in the high level and the potential in the node N.sub.14 is placed in the low level so that this circuit is reset, the transistor Q.sub.41 is placed in the off state, the transistor Q.sub.42 is placed in the on state and the signal OBD is placed in the low level state. After the reset is completed, at the time t.sub.2, the signal DBD is placed in the high level and the signal CDD is placed in the low level so that this circuit commences operation.

When node N.sub.14 is placed in the high level and the node N.sub.12 is placed in the low level, the transistor Q.sub.41 is placed in the on state, the transistor Q.sub.42 is in the off state, the signal OBD is placed in the high level so that the output buffer 19b is driven.

The circuit which includes the transistors Q.sub.43 .about.Q.sub.52 and the resistor R.sub.61 is the circuit for forming the signal DBR and the timing chart of this circuit is shown in FIG. 9B. This circuit is reset by the signal CDD. After this circuit is reset, when the signal OBD is placed in the high level, the signal DBR is also placed in the high level. Thereafter, the node N.sub.18 is placed in the high level by the timing circuit including the transistors Q.sub.43 .about.Q.sub.48 and the resistor R.sub.61. Therefore, the transistor Q.sub.50 is placed in the on state, the node N.sub.19 is placed in the low level and the transistor Q.sub.51 is placed in the off state. On the other hand, the transistor Q.sub.52 is placed in the on state so that the signal DBR is placed at the low level via the transistor Q.sub.52. FIG. 8C is a circuit of the output buffer 19b, symbols Q.sub.61 through Q.sub.68 denotes MOS transistors and symbols N.sub.21 through N.sub. 24 denote nodes or potentials at the nodes. The time chart of the output buffer 19b is shown in FIG. 9C. At the time t.sub.1, the node N.sub.23 is placed at the high level and the transistor Q.sub.67 is placed in the on state therefore the high level is output at the output terminal D.sub.out. At the time t.sub.2, the signal OBD is placed in the low level, the potential of node N.sub.23 is discharged via the transistor Q.sub.63 to the signal OBD, the transistor Q.sub.67 is placed in the off state and is reset. At the time t.sub.3, the potential levels of the nodes N.sub.21 and N.sub.22 are determined by the signal RD, RD. That is, one of the nodes N.sub.21 and N.sub.22 is placed at the high level and another of them is placed in the low level. In the example shown in FIG. 9C, the node N.sub.22 is in the high level and the node N.sub.21 is in the low level. Thereafter, when the signal OBD is placed at the high level, as the transistor Q.sub.64 is in the on state, the node N.sub.24 is placed in the high level, so that the transistor Q.sub.63 is placed in the on state. On the other hand, if the signal OBD is placed in the high level, the transistor Q.sub.63 is in the off state, therefore, the node N.sub.23 is maintained at the low level and the transistor Q.sub.67 is in the off state. Therefore, the output D.sub.out is placed in the low level state.

According to the present invention as illustrated in the foregoing, the individual functional blocks or interface circuits (except the output buffer) which have finished the operation, are readily reset by the signals of functional block of the next stage or of the next but one stage. And, the output buffer which maintains last read data on terminal D.sub.out is reset with a signal from the column decoder just before the output buffer commences the new operation. Therefore, it is possible to greatly reduce the cycle time of the dynamic memory and, eventually, to write and read large amounts of data within reduced periods of time.