Semiconductor Integrated Circuit Device Allowing Change of Product Specification and Chip Screening Method Therewith

1. Field of the Invention

This invention relates to a semiconductor integrated circuit device allowing the change of the product specification and a chip screening method therewith.

2. Description of the Related Art

In dynamic RAMs (hereinafter, referred to as DRAMs), the ratio of the refresh time T to the refresh cycle R is T/R=15.6 .mu.sec. This ratio holds true for each generation of DRAMs: for example, 8 msec/512 cycles for the 1Mbit DRAM generation, and 16 msec/1024 cycles (hereinafter, 1024 cycles are referred to as 1-kcycles) for the 4Mbit DRAM generation.

For the 16Mbit DRAM generation and later, the relationship should be 32 msec/2048 cycles (hereinafter, 2048 cycles are referred to as 2-kcycles). To reduce power consumption, prevent heat generation, and make the active current smaller, however, it is necessary to increase the number of refresh cycles to decrease the number of cell arrays to be activated at the same time. For example, the number of refresh cycles is increased to 4096 cycles (hereinafter, referred to as 4-kcycles).

Additionally, there is a need to reduce the number of refresh cycles in order to manufacture multi-bit symmetrical address products. For example, 1-kcycles are used. To prevent the chip size from becoming larger and to ensure the sensitivity (C.sub.B /C.sub.S where C.sub.B is the bit-line capacity and C.sub.S is the cell capacity), the number of cells per bit-line cannot be changed from the present value (for examine, 128 cells per bit-line), so that it is natural to change the number of refresh cycles.

Changing the number of refresh cycles means that the chip must be redesigned each time the number is changed. This imposes a heavy burden on the circuit designing personnel, resulting in reduced development efficiciency.

More diversification of products requires factories to produce a variety of products simultaneously, reducing the production efficiency.

Additionally, the conventional chip screening test only rejects defective products. In this test, chips that fail to come up to the passing mark set for each product are judged to be unacceptable, and are discarded. Because of this, the conventional chip screening test has contributed to a poorer product yield.

The object of the present invention is to provide a semiconductor integrated circuit device which is a solution to the problem that the diversification of products reduces development and production efficiencies, or which allows the product diversification without the sacrifice of development and production efficiencies.

Another object of the present invention is to provide a chip screening method capable of improving the product yield.

The foregoing object is accomplished by providing a semiconductor integrated circuit device comprising: an integrated circuit section containing a first circuit section having a first function and a second circuit section having a second function; an active signal generator section for producing an active signal for activating the first circuit section or the second circuit section; receiving means for taking in a decision signal for determining the product specification; a switching signal generator section, connected to the receiving means, for producing a switching signal for changing the product specification based on the decision signal; and switching means which, based on the switching signal, changes the supply of the active signal to either the first circuit section or to the second circuit section.

With the present invention, a decision signal to determine the product specification is produced, and based on this signal, the function of the integrated circuit section is modified to meet the product specification, thereby making it possible to produce more than one type of product from a single product. This makes it unnecessary for the designing personnel to design the respective circuits to meet product specifications (or product types), increasing development efficiency. This approach also allows various types of products to share almost all manufacturing processes, improving production efficiency.

Further, the second object is accomplished by providing a chip screening method comprising: the chip screening step of selecting semiconductor chips, which includes a select test for determining whether semiconductor chips are acceptable or not, and a pause test for checking the memory cell for the charge retaining characteristics; and the product specification switching step of changing the product specification of the chips based on the result of the pause test.

In the chip screening test, by changing the product specification to that of the chip corresponding to the pause time based on the result of the pause test in the screening step, chips that would be unacceptable in the conventional test can be saved. This prevents the yield especially in the screening test from decreasing, and consequently improves the product yield.

Additional objects and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objects and advantages of the invention may be realized and obtained by means of the instrumentalities and combinations particularly pointed out in the appended claims.

The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate presently preferred embodiments of the invention, and together with the general description given above and the detailed description of the preferred embodiments given below, serve to explain the principles of the invention.

FIG. 1 is a block diagram of a DRAM according to a first embodiment of the present invention;

FIG. 2 is a circuit diagram of the product specification deciding circuit of FIG. 1;

FIG. 3 is a circuit diagram of another example of the receiving section of FIG. 2;

FIG. 4 is a block diagram of the counter circuit of FIG. 1;

FIGS. 5A and 5C are circuit diagrams of the counters of FIG. 4;

FIGS. 6A and 6B are circuit diagrams of the word-line step-up section of FIG. 1;

FIG. 7 is a circuit diagram of another example of the word-line step-up section of FIG. 1;

FIG. 8 is a circuit diagram of the X2-decoder of FIG. 1;

FIG. 9 is a circuit diagram of the I/O sense amplifier group and I/O sense amplifier control circuit of FIG. 1;

FIG. 10 is a block diagram of a DRAM according to a second embodiment of the present invention;

FIG. 11 is a circuit diagram of the product specification deciding circuit of FIG. 10;

FIG. 12 is a block diagram of a DRAM according to a third embodiment of the present invention;

FIG. 13 is a circuit diagram of the product specification deciding circuit of FIG. 12;

FIG. 14 is a block diagram of a DRAM according to a fourth embodiment of the present invention;

FIG. 15 is a circuit diagram of the receiving section and switching signal generator section of FIG. 14;

FIG. 16 is a circuit diagram of the address switching section of FIG. 14;

FIGS. 17A to 17C are circuit diagrams of the X-address buffer group of FIG. 14;

FIGS. 18A and 18B are circuit diagrams of the Y-address buffer group of FIG. 14;

FIGS. 19A to 19C are circuit diagrams of the counter circuit group of FIG. 14;

FIG. 20 is a circuit diagram of the word-line step-up section of FIG. 14;

FIG. 21 shows the logic of VR1K, VR2K, R1K, R2K and R4K for each refresh cycle;

FIG. 22 shows the destinations of outputs A and B for each refresh cycle;

FIGS. 23A to23C show address allocation for each refresh cycle;

FIG. 24 is a block diagram of the I/O sense amplifier group of FIG. 1;

FIG. 25 is a block diagram of the I/O sense amplifier group of FIG. 1;

FIG. 26 is a flowchart of a chip screening method according to the present invention;

FIG. 27 is a flowchart of another example of the chip screening method;

FIG. 28 shows the contents of step 2 in FIGS. 26 and 27; and

FIG. 29 is a sectional view of the pad of FIG. 2.

Referring to the accompanying drawings, embodiments of the present invention will be explained. Like parts are indicated by corresponding reference characters throughout all the figures, and repetitious explanation will be omitted.

FIG. 1 is a block diagram of a DRAM according to a first embodiment of the present invention. The DRAM, i.e., the first embodiment, can operate in 2k-refresh cycle mode and 4k-refresh-cycle mode.

As shown in FIG. 1, a memory cell array (hereinafter, referred to as the MCA) 1 is divided into eight sections MCA.sub.0 to MCA.sub.7. An X-address buffer group 3, which receives an address input signal A.sub.in, produces a plurality of X-address signals. The X-address signals are set to first X-addresses X.sub.0 to X.sub.8 and second X-addresses X.sub.9 and X.sub.10 for division operation of MCA.sub.0 to MCA.sub.7, and to a third X-address X.sub.11 for changing the product specification. An X1-decoder 5, which is supplied with the first X-addresses X.sub.0 to X.sub.8, decodes the first addresses X.sub.0 to X.sub.8 to produce a signal for selecting a word-line (a row) of the MCA. An X2-decoder 7 is supplied with the second X-addresses X.sub.9 and X.sub.10 and also with the third X-address X.sub.11x via an address switching section 9. When the 2-kcycle DRAM is selected, the X2-decoder 7 decodes the second X-addresses X.sub.9 and X.sub.10 to produce a signal for simultaneously selecting one array from MCA.sub.0 to MCA.sub.3 and one array from MCA.sub.4 to MCA.sub.7, a signal for selecting sense amplifiers 11.sub.0 to 11.sub.3, and a signal for selecting I/O sense amplifier groups 13.sub.0 to 13.sub.3. When the 4-kcycle DRAM is selected, the X2-decoder 7 decodes the second X-addresses X.sub.9 and X.sub.10 and the third X-address X.sub.11 to produce a signal for selecting one array from MCA.sub.0 to MCA.sub.7, a signal for selecting sense amplifiers 11.sub.0 to 11.sub.3, and a signal for selecting the I/O sense amplifier groups 13.sub.0 to 13.sub.3. In FIG. 1, blocks indicated by reference numerals 15.sub.0 to 15.sub.3 are word-line driving circuits, and blocks indicated by reference numerals 17.sub.0 to 17.sub.3 are sense amplifier driving circuits. A Y-address buffer group 19, which receives the address input signal A.sub.in, generates a plurality of Y-address signals. The Y-address signals are set to first Y-addresses Y.sub.1 to Y.sub.11 and a second Y-address Y.sub.0. A Y1-decoder 21, which is supplied with the first Y-addresses Y.sub.1 to Y.sub.11 decodes the first Y-addresses Y.sub.1 to Y.sub.11 to produce a signal for selecting a bit line (a column) of the MCA. A Y2-decoder 23 decodes the second Y-address Y.sub.0 to generate a signal for selecting, for example, one of the I/O sense amplifiers contained in the I/O sense amplifier group 13.

The DRAM of FIG. 1 is provided with a product specification deciding section 25 for deciding the product specification semipermanently. The product specification deciding section 25 is composed of a receiving section 27 that receives a product specification decision signal SDS to decide the product specification semipermanently, a switching signal generator section 29, connected to the receiving section 27, for producing internal switching signals .phi.2 and .phi.4 to change product specifications based on the signal SDS, and an address signal switching section 9 for selecting the destination of the address signal based on the signals .phi.2 and .phi.4.

The operation of the product specification deciding section 25 will be explained.

When the product specification decision signal SDS specifies the 2-kcycle refresh product (mode), the switching signal generator section 29 produces a 2-kcycle refresh product (mode), the switching signal generator section 29 produces a 2-kcycle refresh product (mode) switching signal .phi.2, and supplies it to the address signal switching section 9 and I/O sense amplifier control circuit 31. The address signal switching section 9, based on the signal .phi.2, changes the third address X.sub.11 to address X.sub.11Y and supplies the resulting signal to the I/O sense amplifier control circuit 31.

When the product specification decision signal SDS specifies the 4-kcycle refresh product (mode), the switching signal generator section 29, based on the signal SDS, produces a 4-kcycle refresh product (mode) switching signal .phi.4, and supplies it to the address signal switching section 9 and X2-decoder 7. The address signal switching section 9 changes the third address X.sub.11 to address X.sub.11x based on the signal .phi.4 and supplies the resulting signal to the X2-decoder 7. The signals .phi.2 and .phi.4 are, for example, complementary to each other. The switching signal generator section 29 supplies the inversion in level of signal .phi.2 to the I/O sense amplifier control circuit 31.

The data read operation of the 2-kcycle refresh memory product and the 4-kcycle refresh product (mode) will be explained.

In the case of the 2-kcycle refresh product (mode), the I/O sense amplifier control circuit 31 is supplied with address X.sub.11Y, which activates the former. The control circuit 31 produces a signal for selecting either a pair of I/O sense amplifiers 13.sub.0 and 13.sub.1 or a pair of I/O sense amplifiers 13.sub.2 and 13.sub.3. The X2-decoder 7 produces a signal for simultaneously selecting one array from MCA.sub.0 to MCA.sub.3 and one array from MCA.sub.4 to MCA.sub.7. The I/O sense amplifier group that finally supplies the data is one selected by the X2-decoder 7 and control circuit 31. The reading of data is done by causing the Y1-decoder 21 to decode the first Y-address produced at the Y-address buffer group 19, amplifying the information from the memory cell at the I/O sense amplifier group that finally supplies the data, and supplying the output signal Dout from the data output circuit 33. In FIG. 1, a block indicated by numeral 35 is a data input circuit to which the input signal Din is supplied.

In the case of the 4-kcycle refresh product (mode), address X.sub.11X is supplied to the X2-decoder 7 instead of the I/O sense amplifier control circuit 31. The X2-decoder 7 then produces a signal for activating only one array of MCA.sub.0 to MCA.sub.7. The control signal 31 receives the inversion in level of signal .phi.2, and based on the inverted signal, produces a signal for selecting either a pair of I/O sense amplifiers 13.sub.0 and 13.sub.1 or a pair of I/O sense amplifiers 13.sub.2 and 13.sub.3. The I/O sense amplifier 13 finally activated is one selected by the X2-decoder 7 and control circuit 31.

As described above, a semiconductor integrated circuit device thus constructed enables a single chip to deal with different refresh-cycles by switching the third X-address X.sub.11 to either X.sub.11X or X.sub.11Y at the switching section 9.

Refresh operation is performed by selecting a word-line and at the same time, by operating the sense amplifiers 11.sub.0 to 11.sub.3.

The DRAM of FIG. 1 is provided with a counter refresh circuit group 37, which contains a counter circuit 39. The counter circuit 39 is supplied with a signal CTRS for commanding the count start and switching signals .phi.2 and .phi.4. The counter circuit 39, based on the signal CTRS, supplies counter output signals C.sub.0 to C.sub.11 that count up X-addresses X.sub.0 to X.sub.11 in sequence, and based on the signals .phi.2 and .phi.4, changes the number of output signals C.sub.0 to C.sub.11. This is done to make the number of X-addresses equal to the number of counter output signals, because the 2-kcycle product (mode) differs from the 4-kcycle product (mode) in the number of X-addresses supplied to the row decoder (X1-decoder 5 and X2-decoder 7). In this embodiment, when the switching signal .phi.2 is supplied, the counter circuit 39 will not supply signal C.sub.11. This is because the third X-address Xi, is ignored since in the case of the 2-kcycle product (mode), the third X-address X.sub.11 is not supplied to the row decoder (X1-decoder 5 and X2-decoder 7). When the switching signal .phi.4 is supplied (or when the level of switching signal .phi.2 is reversed and supplied), the counter circuit 39 will supply signal C.sub.11.

The DRAM of FIG. 1 is provided with a word-line boosting section 41, to which switching signals .phi.2 and .phi.4 and boosting signal .phi.WL are supplied. The word-line boosting section 41 raises the word-line voltage based on signal .phi.WL. In FIG. 1, numeral 43 indicates the boosting line to which a boosting voltage is supplied. In the present invention, the word-line boosting capacitance is also changed on the basis of signals .phi.2 and .phi.4. This is done to optimize the level of the word-line boosting capacitance according to a change in word-line load capacitance, since the number of word lines activated at a time in the 2-kcycle product differs from that in the 4-kcycle products. In the case of 2-kcycle products, because two MCAs are selected, this increases the number of word lines activated, making the load capacitance larger. To compensate for the increase in the load capacitance, the word-line boosting section 41 increases the word-line boosting capacitance based on signal .phi.2 in the case of the 2-kcycle product. When signal .phi.4 is supplied (or when the level of switching signal .phi.2 is inverted and supplied; in the case of 4-kcycle product), the word-line boosting section 41 reduces the word-line boosting capacitance more than in the 2-kcycle product.

The peripheral circuitry of the FIG. 1 DRAM contains a /RAS (hereinafter, / is used as a symbol indicating an inverted signal) circuit group 45, a /CAS circuit group 47, and a /WE circuit group 49. The details of these circuits will be omitted in this specification.

FIG. 2 is a circuit diagram showing a concrete construction of the product specification deciding section 25.

As shown in FIG. 2, the receiving section 27 is composed of a pad P connected to the output terminal 51, and a resistance one end of which is connected to the junction point of the output terminal 51 and pad P and the other end of which is connected to the ground GND. This section 27 allows the output terminal 51 to be set to either a H (high) level or a L (low) level depending on whether a wire applied with a high potential VCC is bonded to the pad P (the decision signal SDS is in the H-level) or not (the signal SDS is in the L-level). The output terminal 51 is connected to the input terminal 53 of the switching signal generator section 29.

The switching signal generator section 29 is made up of a first inverter 55 whose input is connected to the input terminal 53, and a second inverter 57 whose input is connected to the output of the first inverter 55. The output of the inverter 55 is extracted as a first refresh switching signal .phi.2, and the output of the inverter 57 is extracted as a second refresh switching signal .phi.4.

The address switching section 9 is composed of switches (transfer gates) 59.sub.1 to 59.sub.4 consisting of n-channel MOSFET (hereinafter, referred to as NMOS) and p-channel MOSFET (hereinafter, referred to as PMOS) whose gates are supplied with switching signals .phi.2 or .phi.4. The X-address buffer group 3 supplies an address signal A.sub.11R (X.sub.11) and its inverted signal /A.sub.11R (/X.sub.11). The address signal A.sub.11R (X.sub.11) is supplied to one end of each of switches 59.sub.1 and 59.sub.2. The other end of switch 59.sub.1 is connected to the X2-decoder 7, and the other end of the switch 59.sub.2 is connected to the I/O sense amplifier control circuit 31. The inverted signal /A.sub.11R (/X.sub.11) is supplied to one end of each of switches 59.sub.3 and 59.sub.4. The other end of switch 59.sub.3 is connected to X2-decoder 7, and the other end of switch 59.sub.4 is connected to the I/O sense amplifier control circuit 31.

The gate of each of the PMOS of switch 59.sub.1 , NMOS of switch 59.sub.2, PMOS of switch 59.sub.3, and NMOS of switch 59.sub.4 is all connected to the output of the inverter 55. The gate of each of the NMOS of switch 59.sub.1, PMOS of switch 59.sub.2, NMOS of switch 59.sub.3, and PMOS of switch 59.sub.4 is all connected to the output of the inverter 57.

Connecting this way allows either a pair of switches 59.sub.1 and 59.sub.3 or a pair of switches 59.sub.2 and 59.sub.4 to be selected and operated. For example, when the output of inverter 55 is in the H-level and the output of inverter 57 is in the L-level (in the case of the 2-kcycle refresh product), the switches 59.sub.2 and 59.sub.4 turn on, and address signal A.sub.11R and its inverted signal /A.sub.11R are supplied as addresses X.sub.11Y and /X.sub.11Y to the I/O sense amplifier control circuit 31.

Contrarily, when the output of inverter 55 is in the L-level and the output of inverter 57 is in the H level (in the case of the 4-kcycle refresh product), the switches 59.sub.1 and 59.sub.3 turn on, and address signal A.sub.11R and its inverted signal /A.sub.11R are supplied as addresses X.sub.11X and /X.sub.11X to the X2-decoder 7.

As noted above, the product specification deciding section 25, depending on whether to bond a wire applied with a high voltage VCC to the pad P or not, switches address signal A.sub.11R and its inverted signal /A.sub.11R either to the X2-decoder 7 or to the I/O sense amplifier control circuit 31.

FIG. 3 is a circuit diagram showing another construction of the receiving section 27.

The receiving section 27 of FIG. 2 may be constructed as shown in FIG. 3. Specifically, one end of the resistance R is connected to the high potential VCC, and the other end of the resistance R is connected to one end of the fuse F, the other end of which is connected to the ground GND. The junction point of the resistance R and fuse F is connected to the output terminal 51.

In the receiving section 27 thus constructed, cutting the fuse F enables the output terminal 51 to be set to the H-level, and uncutting the fuse F allows the output terminal 51 to be set to the L-level. The receiving section 27 of FIG. 3 operates in the same manner as that of FIG. 2.

FIG. 4 is a block diagram of the counter circuit 39 of FIG. 1.

As shown in FIG. 4, the counter circuit 39 is composed of counters 61.sub.0 to 61.sub.11. The least-significant counter 61.sub.0 is supplied with the signal CTRS commanding the count start and its inverted signal BCTRS. The counter 61.sub.0, based on the signal CTRS and its inverted signal BCTRS, supplies a counter output signal C.sub.0 and its inverted signal BC.sub.0. The counter 61.sub.1 in the next stage is supplied with the output (signal C.sub.0 and its inverted signal BC.sub.0) of the counter 61.sub.0 in the preceding stage. The counter 61.sub.1, based on the signal C.sub.0 and its inverted signal BC.sub.0, supplies a counter output signal C.sub.1 and its inverted signal BC.sub.1. In this way, counters 61.sub.1 to 61.sub.11 take in the outputs of the preceding stages, respectively, and based on the signals taken in, supply signals C.sub.1 to C.sub.11 and their inverted signals BC.sub.1 to BC.sub.11 in sequence. The most-significant counter 61.sub.11 is supplied with the output (signal C.sub.10 and its inverted signal BC.sub.10) of the counter 61.sub.10 in the preceding stage (not shown) and switching signal .phi.4. The counter 61.sub.11, only when, for example, supplied with the H-level switching signal .phi.4 (in the case of the 4-kcycle refresh product), supplies counter output signal C.sub.11 and its inverted signal BC.sub.11 on the basis of signal C.sub.10 and its inverted signal BC.sub.10. The counter 61.sub.11, when, for example, supplied with the L-level switching signal .phi.4 (in the case of the 2-kcycle refresh product), supplies neither signal C.sub.11 nor its inverted signal BC.sub.11. Thus, for the 2-kcycle refresh product, the output of the counter 61.sub.11 is ignored.

FIGS. 5A to 5C are a circuit diagram showing a concrete construction of the counters of FIG. 4.

The circuit configuration of each of counters 61.sub.0 to 61.sub.10 is the same, so that only counters 61.sub.0 and 61.sub.1 and the most-significant counter 61.sub.11 will be described.

FIGS. 5A and 5B are circuit diagrams of counters 61.sub.0 and 61.sub.1, respectively.

As shown in FIG. 5A, the output of the clocked inverter 63.sub.0 is connected to the input of the inverter 65.sub.0 (node a1). The output of inverter 65.sub.0 is connected to the gate of each of PMOS 67.sub.0 and NMOS 69.sub.0. The drain of PMOS 67.sub.0 is connected to that of NMOS 69.sub.0 (node a2). The source of PMOS 67.sub.0 is connected to the drain of PMOS 71.sub.0, and the source of PMOS 71.sub.0 is connected to a high potential power supply. The gate of PMOS 71.sub.0 is supplied with signal CTRS. The source of NMOS 69.sub.0 is connected to the drain of NMOS 73.sub.0, and the source of NMOS 73.sub.0 is connected to a low potential power supply (for example, the ground). The gate of NMOS 73.sub.0 is supplied with the inverted signal BCTRS. Node a2 is connected to node a1 as well as to the input of the clocked inverter 75.sub.0, which is driven by the clock opposite in phase to that of the clocked inverter 63.sub.0. The output of the clocked inverter 75.sub.0 is connected to the input of the inverter 77.sub.0 (node a3). The output of the inverter 77.sub.0 is connected to the gate of each of PMOS 79.sub.0 and NMOS 81.sub.0 (node a4). The drain of PMOS 79.sub.0 is connected to that of NMOS 81.sub.0 (node a5). The source of PMOS 79.sub.0 is connected to the drain of PMOS 83.sub.0, and the source of PMOS 83.sub.0 is connected to a high potential power supply. The gate of PMOS 83.sub.0 is supplied with the inverted signal BCTRS. The source of NMOS 81.sub.0 is connected to the drain of NMOS 85.sub.0, whose source is connected to a low potential power supply (for example, the ground). The gate of NMOS 85.sub.0 is supplied with the signal CTRS. Node a5 is connected to node a3. Node a4 is connected to the counter output signal terminal Cj (C.sub.0) (node a6). Node a6 is connected to the input of the inverter 87.sub.0 (node a7). The output of inverter 87.sub.0 is connected to the inverted counter output signal terminal BCj (BC.sub.0). Node a7 is connected to the input of inverter 63.sub.0. Explanation of FIG. 5B will be omitted. The construction of FIG. 5B is almost the same as that of FIG. 5A except for input signals (Cj-1, BCj-1) and output signals (Cj, BCj).

The most-significant counter 61.sub.11 will be explained.

As shown in FIG. 5C, node a2 is connected to the gate of PMOS 89.sub.11 (node a8) as well as to the gate of NMOS 91.sub.11. Node a8 is connected to node a1. The drain of PMOS 89.sub.11 is connected to the source of PMOS 93.sub.11. The source of PMOS 89.sub.11 is connected to a high potential power supply. The gate of PMOS 93.sub.11 is connected to signal CJ-1 (C.sub.10). The drain of NMOS 91.sub.11 is connected to the source of NMOS 95.sub.11, which is also connected to the drain of NMOS 97.sub.11. The gate of NMOS 95.sub.11 is supplied with the inverted signal BCj-1 (C.sub.10). The drain of PMOS 93.sub.11 is connected to that of NMOS 95.sub.11 (node a9). Node a9 is connected to the drain of PMOS 99.sub.11 whose source is connected to a high potential power supply. The gate of each of PMOS 99.sub.11 and NMOS 97.sub.11 is supplied with switching signal .phi.4. Node a9 is connected to node a3.

The operation of the counter of FIG. 5 will be explained.

It is assumed that the first stage counter 61.sub.0 is supplied with signal Cj-1 (CTRS) and inverted signal BCj-1 (BCTRS), and that the clocked inverter 63.sub.0 and the clocked inverter 101.sub.0 made up of PMOS 79.sub.0 and PMOS 83.sub.0 and NMOS 81.sub.0 and NMOS 85.sub.0 are turned on. In this state, the clocked inverter 75.sub.0 and the clocked inverter 103.sub.0 made up of PMOS 67.sub.0 and PMOS 71.sub.0 and NMOS 69.sub.0 and NMOS 73.sub.0 are in the off state because they are supplied with the clock opposite in phase to that of the clocked inverter 63.sub.0. As a result, a latch circuit composed of the inverter 77.sub.0 and clocked inverter 101.sub.0 latches a signal that brings node a4 to the H-level. This allows the counter output signal terminal Cj to supply the H-level signal (C.sub.0), and the inverted counter output signal terminal BCj to supply the L-level signal (BC.sub.0). When the level of the clock signal is inverted, the clocked inverters 63.sub.0 and 101.sub.0 are turned off and the clocked invertors 75.sub.0 and 103.sub.0 are turned on. As a result, a latch circuit composed of the inverter 65.sub.0 and clocked inverter 103.sub.0 latches a signal that brings node a2 to the L-level. When node a2 is in the L-level, the clocked inverter 75.sub.0 supplies the H-level signal, bringing node a4 to the L-level. Therefore, the counter output signal terminal Cj supplies the L-level signal (C.sub.0) opposite in level to that of the signal described above, and the inverted counter output signal terminal BCj supplies the H-level signal (BC.sub.0) whose signal level has been inverted. The next-stage counter 61.sub.1 is supplied with the output signals C.sub.0 and BC.sub.0 and driven by them. The subsequent counters 61.sub.2 to 61.sub.10 operate the same way. The eleventh-stage counter 61.sub.10 supplies signals C.sub.10 and BC.sub.10, which are used to drive the final counter 61.sub.11. In the counter 61.sub.11, a low potential is supplied via NMOS 97.sub.11 to the clocked inverter 75.sub.11 made up of PMOS 89.sub.11, PMOS 93.sub.11, NMOS 91.sub.11, and NMOS 95.sub.11. The gate of NMOS 97.sub.11 is supplied with the switching signal .phi.4. Because the L-level switching signal turns off NMOS 97.sub.11, the clocked inverter 75.sub.11 does not operate. Thus, the counter 61.sub.11 supplies effective counter output signal C.sub.j (C.sub.11) and inverted output signal BCj (BC.sub.11) only when the switching signal is in the H-level.

FIGS. 6A and 6B are circuit diagrams showing a concrete construction of the word-line boosting section 41 of FIG. 1.

As shown in FIG. 6A, the word-line boosting section 41 contains a first boosting capacitor 105.sub.1 and second boosting capacitor 105.sub.2. One electrode of each of the first and second boosting capacitors 105.sub.1 and 105.sub.2 is connected to the boosting line 43. The line 43 is connected to the boosting driving circuit 15.sub.0 to 15.sub.7 shown in FIG. 1. The other electrode of capacitor 105.sub.1 is connected to the output of the first word-line boosting circuit 107.sub.1, and the other electrode of capacitor 105.sub.2 is connected to the output of the second word-line boosting circuit 107.sub.2. The input of the first word-line boosting circuit 107.sub.1 is supplied with a boosting signal .phi.WL. The input of the second word-line boosting circuit 107.sub.2 is connected to the output of the AND gate (logical product gate) 109. The input of AND gate 109 is supplied with the signal .phi.WL and switching signal .phi.2. Each of the boosting circuits 107.sub.1 and 107.sub.2 is composed of two inverters connected in series between the input and output.

The operation of the word-line boosting section 41 of FIG. 6 will be explained. When both the boosting signal .phi.WL and switching signal .phi.2 are in the H-level (in the case of the 2-kcycle product), both boosting circuits 107.sub.1 and 107.sub.2 are activated. When the switching signal .phi.2 is in the L-level (in the case of the 4-kcycle product), only the boosting circuit 107.sub.1 is activated. Thus, the boosting section 41 of the 2-kcycle product supplies a higher boosting capacitance than that of the 4-kcycle product.

As shown in FIG. 6B, the step-up section 41 may be made up of the boosting circuit 107.sub.2 connected between the input and output with the input being connected to a NAND gate 111. The boosting section 41 of FIG. 6B operates in the same manner as the boosting section 41 of FIG. 6A.

FIG. 7 is a block diagram showing another construction of the word-line boosting section 41.

As shown in FIG. 7, the boosting line 43 connected to the second boosting capacitor 105.sub.2 may be prepared as a mask option. Specifically, in the manufacturing processes, the conducting layer patterning of the boosting line 43 may be designed to allow selection of a mask with the pattern of boosting line 43 connected only to the first boosting capacitor 105.sub.1 or a mask with the pattern of boosting line 43' connected to the second boosting capacitor 105.sub.2 in addition to that of the first one.

FIG. 8 is a circuit diagram showing a concrete construction of the X2-decoder 7 of FIG. 1.

AND gates 113.sub.0 to 113.sub.7 are provided as shown in FIG. 8. The inputs of AND gates 113.sub.0 to 113.sub.7 are supplied with the second addresses X.sub.9 (/X.sub.9) and X.sub.10 (/X.sub.10) and the third address X.sub.11X (/X.sub.11X) in a different combination. The third address inputs of AND gates 113.sub.0 to 113.sub.7 are connected to either the sources or drains of PMOS 115.sub.0 to PMOS 115.sub.7. The gates of PMOS 115.sub.0 to PMOS 115.sub.7 are supplied with the switching signal .phi.4. The outputs CBS0 to CBS7 of AND gates 113.sub.0 to 113.sub.7 are extracted as cell array block select signals.

The operation of X2-decoder 7 will be explained. When the switching signal .phi.4 is in the L-level (in the case of the 2-kcycle product), PMOS 115.sub.0 to PMOS 115.sub.7 are each turned on, causing the third address input to remain at the H-level. Therefore, the third X-address input (X.sub.11X and /X.sub.11X) is ignored. When the switching signal .phi.4 is in the H-level (in the case of the 4-kcycle product), PMOS 115.sub.0 to PMOS 115.sub.7 are each turned off, activating the third X-address input. As a result, AND gates 113.sub.0 to 113.sub.7 take in addresses X.sub.11X and /X.sub.11X.

FIG. 9 is a circuit diagram showing a concrete construction of the I/O sense amplifier group 13 and I/O sense amplifier control circuit 31 shown in FIG. 1.

As shown in FIG. 9, the I/O sense amplifier control circuit 31 contains AND gates 117.sub.0 and 117.sub.1. Each of AND gates 117.sub.0 and 117.sub.1 is supplied with the I/O sense timing signal .phi..sub.IOS and address X.sub.11Y and /X.sub.11Y. The third X-address input of each of AND gates 117.sub.0 and 117.sub.1 is connected to the sources or drains of PMOS 119.sub.0 and PMOS 119.sub.1. The gates of PMOS 119.sub.0 and PMOS 119.sub.1 are supplied with the switching signal .phi.2. The outputs .phi..sub.S01 and .phi..sub.S23 of AND gates 117.sub.0 and 117.sub.1 are extracted as the I/O sense amplifier group select signals to select the I/O sense amplifier groups 13.sub.0 to 13.sub.3.

The operation of the I/O sense amplifier control circuit 31 will be explained.

When the switching signal .phi.2 is in the L-level (in the case of the 4-kcycle product), PMOS 119.sub.0 to PMOS 119.sub.7 are each turned on, causing the third X-address input to remain at the L-level. Therefore, the third X-address input (X.sub.11Y and /X.sub.11Y) is ignored. When the switching signal .phi.2 is in the H-level (in the case of the 2-kcycle product), PMOS 119.sub.0 and PMOS 119.sub.1 are each turned off, activating the third X-address input. As a result, AND gates 117.sub.0 to 117.sub.1 take in addresses X.sub.11Y and /X.sub.11Y.

FIG. 9 shows a primary portion of the I/O sense amplifier groups 13.sub.0 to 13.sub.3.

As shown in FIG. 9, the I/O sense amplifier groups 13.sub.0 to 13.sub.3 are each made up of OR gates 121.sub.0 to 121.sub.3 and AND gates 123.sub.0 to 123.sub.3. The inputs of OR gates 121.sub.0 to 121.sub.3 are supplied with block select signals CBSO to CBS7. The inputs of AND gates 123.sub.0 to 123.sub.3 are supplied with the outputs of OR gates 121.sub.0 to 121.sub.3, and I/O sense amplifier select signals .phi..sub.S01 and .phi..sub.S23. The outputs of AND gates 123.sub.0 to 123.sub.3 are extracted as I/O sense timing signals .phi..sub.IOS0 and .phi..sub.IOS3.

A second embodiment of the present invention will be explained.

FIG. 10 is a block diagram of a DRAM according to the second embodiment of the present invention. This figure centers especially on the product specification deciding section 25. The DRAM shown in FIG. 1 is a device where the X-address allocating method is the same as the Y-address allocating method, such as a DRAM of a x 1 bit construction. Two types of products with different refresh-cycles can be obtained from a single DRAM of FIG. 1.

In some devices, however, as the refresh cycle changes, the X-address allocation and Y-address allocation change accordingly. They include x4 bit DRAMs, x8 bit DRAMs, and x16 bit DRAMs, or multi-bit DRAMs. In the multi-bit DRAM, as the refresh-cycle changes, the number of X-addresses and that of Y-addresses change. Therefore, to realize several types of products with different refresh-cycles, it is necessary to change the allocation of X-addresses and Y-addresses according to the difference in refresh cycle. A DRAM according to the second embodiment is a device that allows the change of address allocation depending on the difference in refresh-cycle.

FIG. 10 is a block diagram of a x4 bit DRAM. In a DRAM of x4 bits with 2k-refresh cycles, the number of X-addresses is equal to that of Y-addresses, or their address are symmetrical. For example, X-addresses range from X.sub.0 to X.sub.10, and Y-addresses range from Y.sub.0 to Y.sub.10. In a DRAM of x4 bits with 4k-refresh-cycles, the number of X-addresses differ from that of Y-addresses, or their addresses are asymmetrical. For example, X-addresses range from X.sub.0 to X.sub.11, and Y-addresses range from Y.sub.0 to Y.sub.9.

In the DRAM shown in FIG. 10, when the refresh-cycle is set to 4-kcycles, X-address X.sub.11 is changed to address X.sub.11X at the address switching section 9, and then supplied to the X2-decoder 7. At this time, Y-address Y.sub.10 is prevented from being supplied from the Y-address buffer group 19. A detailed description of this will be found in a later embodiment.

When the refresh cycle is set to 2-kcycles, Y-address Y.sub.10 is changed to address X.sub.11Y at the address switching section 9, and then supplied to the I/O sense amplifier control circuit 31. At this time, X-address X.sub.11 is prevented from being supplied from the X-address buffer group 3. As with Y-address Y.sub.10, a detailed description of X-address X.sub.11 will be found in a later embodiment.

FIG. 11 is a circuit diagram of the product specification deciding section 25 of FIG. 10.

As shown in FIG. 11, the address switching section 9 contains switches (transfer gates) 59.sub.1 to 59.sub.4 composed of NMOS and PMOS elements. X-addresses X.sub.11 (A.sub.11R) and /X.sub.11, (/A.sub.11R) are supplied to switches 59.sub.1 and 59.sub.3, respectively. Y-addresses Y.sub.10 (A.sub.10C) and /Y.sub.10 (/A.sub.10C) are supplied to switches 59.sub.2 and 59.sub.4, respectively. Thus, when the switching signal .phi.2 is in the H-level and the switching signal .phi.4 is in the L-level (in the case of the 2-kcycle-refresh product), Y-addresses Y.sub.10 and /Y.sub.10 are supplied as addresses X.sub.11Y and/X.sub.11Y to the I/O sense amplifier control circuit 31 via switches 59.sub.2 and 59.sub.4.

When the switching signal .phi.2 is in the L-level and the switching signal .phi.4 is in the H-level (in the case of the 4-kcycle refresh product), X-addresses X.sub.11 and /X.sub.11 are supplied as addresses X.sub.11X and /X.sub.11X to the X2-decoder 7 via switches 59.sub.1 and 59.sub.3.

The same reasoning may be applied to x8 bit and x16 bit devices.

A third embodiment of the present invention will be explained.

FIG. 12 is a block diagram of a DRAM according to the third embodiment. This figure centers primarily on the product specification deciding section 25. The DRAM of the third embodiment enables the change of refresh cycle as well as bit construction. For example, a single DRAM may be formed into four types of products: a x 1 bit product at 2-kcycles, a x 1 bit product at 4-kcycles, a x4 bit product at 2-kcycles, and a x4 bit product at 4-kcycles.

As shown in FIG. 12, the address switching section 9, based on the switching signals .phi.2 and .phi.4, supplies address Y.sub.10Y to the column decoder 127.

In the DRAM of FIG. 12, for a x 1 bit construction at 2k-refresh cycles, the address switcing section 9, based on the switching signals .phi.2 and .phi.4, changes X-address signal X.sub.11 to address Y.sub.10Y to supply the latter to the column decoder 127.

For a x1bit construction at 4k-refresh cycles, the address switching section 9, based on the switching signals .phi.2 and .phi.4, changes Y-address signal Y.sub.10 to address Y.sub.10Y to supply address Y.sub.10Y to the column decoder 127.

For a x4 bit construction at 2k-refresh cycles and a x4 bit construction at 4k-refresh cycles, the address switching section 9 is prevented from supplying address Y.sub.10Y. An alternative to this is to connect between the address switching section 9 and the column decoder 127 a circuit that ignores address Y.sub.10Y based on the signal specifying a x4 bit construction.

In this way, by constructing the address switching section 9 so that for a x1 bit construction, address Y.sub.10Y may be produced from X-address or Y-address based on the switching signals .phi.2 and .phi.4, while for a x4 bit construction, address Y.sub.10Y may be ignored independently of the switching signals .phi.2 and .phi.4, it is possible to realize a DRAM that enables not only the change of refresh cycle but also that of bit construction.

FIG. 13 is a circuit diagram of the product specification deciding section 25 of FIG. 12.

As shown in FIG. 13, the address switching section 9 contains switches (transfer gates) 59.sub.1 to 59.sub.4 and switches 129.sub.1 to 129.sub.4 composed of NMOS and PMOS elements. X-address X.sub.11 (A.sub.11R) is supplied to switches 59.sub.1 to 129.sub.1. Similarly, the inverted X-address /X.sub.11 (/A.sub.11R) is supplied to switches 59.sub.3 and 129.sub.3 ; Y-address Y10 (A.sub.10C) is supplied to switches 59.sub.2 to 129.sub.2 ; and the inverted /Y.sub.10 (/A.sub.10C) is supplied to switches 59.sub.4 and 129.sub.4. The switching signal .phi.4 is supplied to the gate of each of the NMOS of switch 59.sub.1 , the PMOS of switch 59.sub.2, the NMOS of switch 59.sub.3, the PMOS of switch 59.sub.4, the PMOS of switch 129.sub.1, the NMOS of switch 129.sub.2, the PMOS of switch 129.sub.3, and the NMOS of switch 129.sub.4. The switching signal .phi.2 is supplied to the gate of each of the PMOS of switch 59.sub.1, the NMOS of switch 59.sub.2, the PMOS of switch 59.sub.3, the NMOS of switch 59.sub.4, the NMOS of switch 129.sub.1, the PMOS of switch 129.sub.2, the NMOS of switch 129.sub.3, and the PMOS of switch 129.sub.4.

With the product specification deciding section 25 of the above-described construction, when the switching signal .phi.2 is in the H-level and the switching signal .phi.4 is in the L-level (in the case of the 2-kcycle refresh product of x1 bits), switches 59.sub.2 and 59.sub.4 turn on, so that Y-addresses Y.sub.10 and /Y.sub.10 are supplied to the sense amplifier control circuit 31 via switches 59.sub.2 and 59.sub.4. Further, because switches 129.sub.1 and 129.sub.3 turn on, so that X-addresses X.sub.11 and /X.sub.11 are supplied to the column decoder 127 via switches 129.sub.1 and 129.sub.3.

When the switching signal .phi.2 is in the L-level and the switching signal .phi.4 is in the H-level (in the case of 4-kcycle refresh product of x1 bits), switches 59.sub.1 and 59.sub.3 turn on, so that X-addresses X.sub.11 and /X.sub.11 are supplied to the X2-decoder 7 via switches 59.sub.1 and 59.sub.3. Further, because switches 129.sub.2 and 129.sub.4 turn on, so that Y-addresses Y.sub.10 and /Y.sub.10 are supplied to the column decoder 127 via switches 129.sub.2 and 129.sub.4.

Between the address switching section 9 and column decoder is connected a circuit (not shown) that ignores addresses Y.sub.10Y and /Y.sub.10Y based on the signal specifying a x4 bit construction. To select a x4 bit construction, this circuit is used to prevent addresses Y.sub.10Y and /Y.sub.10Y from being supplied to the column decoder 127.

A fourth embodiment of the present invention will be explained.

FIG. 14 is a block diagram of a DRAM according to the fourth embodiment. This figure centers primarily on the product specification deciding section 25. The DRAM of this embodiment allows the change of refresh cycle to more than two different cycles, for example, any of 1-kcycles, 2-kcycles, and 4-kcycles.

FIG. 15 is a circuit diagram of the receiving section 27 and switching signal generating section 29 of FIG. 14.

As shown in FIG. 15, the receiving section 27 contains two bonding pads P1 and P2. Pad P1 is supplied with a first product specification decision signal VR2K, and pad P2 with a second product specification decision signal VR1K. A first output terminal 200 connected to pad P1 is connected to a first input of a NOR gate 202. A second output terminal 204 connected to pad P2 is connected to a first input of a NAND gate 206. A second input of the NAND gate 206 is connected to bonding pad P3 supplied with the signal x16 determining the bit construction. To select a x16 bit construction, a H-level signal is supplied to pad P3. Supplying a L-level signal to pad P3 allows the formation of the product of a x8 bit construction. The output of a NAND gate 206 is connected to the input of an inverter 208. The output of the inverter 208 is extracted as a first switching signal R1K, and is connected to the second output of the NOR gate 202. The output of the NOR gate 202 is extracted as a third switching signal R4K as well as a second switching signal R2K via an inverter 210. As shown in FIG. 14, among these switching signals R1K, R2K, and R4K, the signals R1K and R4K are supplied to the address switching section 9 and counter circuit 37, while the signals R1K and R2K are supplied to the X-address buffer group 3, Y-address buffer group 19, and word-line step-up section 41.

FIG. 21 shows the logic of VR1K, VR2K, R1K, R2K and R4K for each refresh cycle in the case of the x16 bit product. In the figure, character H indicates a H-level signal, and L a L-level signal.

FIG. 16 is a circuit diagram of the address switching section 9 of FIG. 14.

As shown in FIG. 16, the address switching section 9 contains switches (transfer gates) 212, to 212.sub.4 composed of NMOS and PMOS elements. The switch 212.sub.1 is supplied with Y-address Y8 (A8C). Similarly, the switch 212.sub.2 is supplied with X-address X.sub.11 (A11R); switch 212.sub.3 with Y-address Y.sub.9 (A9C); and switch 212.sub.4 with X-address X.sub.10 (X.sub.10 R). The third switching signal R4K is supplied to the gate of each of the PMOS of switch 212.sub.1 and the NMOS of switch 212.sub.2. The switching signal R4K is also supplied via the inverter 214.sub.1 to the gate of each of the NMOS of switch 212.sub.1 and the PMOS of switch 212.sub.2. The first switching signal R1K is supplied to the gate of each of the NMOS of switch 212.sub.3 and the PMOS of switch 212.sub.4. The switching signal R1K is also supplied via the inverter 214.sub.2 to the gate of each of the PMOS of switch 210.sub.3 and the NMOS of switch 212.sub.4. FIG. 16 shows only the portions to which addresses Y.sub.8, Y.sub.9, X.sub.10, and X.sub.11 are supplied, while omitting the portions to which the inverted addresses /Y.sub.8, /Y.sub.9, /X.sub.10, and /X.sub.11 are supplied. The circuit arrangement of the portions to which the inverted addresses are supplied is the same as that shown in FIG. 16.

With the address switching section 9 of the above construction, when the switching signal R1K is in the H-level and the switching signal R4K is in the L-level (in the case of the 1-kcycle-refresh product), switches 212.sub.1 and 212.sub.3 turn on, which allows Y-addresses Y.sub.8 and Y.sub.9 to be supplied as output signals A and B by way of switches 212.sub.1 and 212.sub.3.

When the switching signal R1K is in the L-level and the switching signal R4K is in the L-level (in the case of the 2-kcycle-refresh product), switches 212.sub.1 and 212.sub.4 turn on, which allows Y-address Y.sub.8 and X-address X.sub.10 to be supplied as output signals A and B by way of switches 212.sub.1 and 212.sub.4.

When the switching signal R1K is in the L-level and the switching signal R4K is in the H-level (in the case of the 4-kcycle-refresh product), switches 212.sub.2 and 212.sub.4 turn on, which allows X-addresses X.sub.10 and X.sub.11 to be supplied as output signals A and B by way of switches 212.sub.2 and 212.sub.4.

FIG. 22 lists the destinations of outputs A and B for each refresh cycle in the case of the x16 bit product. Characters Y8Y, Y9Y, X10X, and X11X in FIG. 22 correspond to those in FIG. 14.

FIG. 17 is a circuit diagram of the X-address buffer group 3 of FIG. 14. FIG. 17A is a circuit diagram of the address generating section that produces addresses A0 to A11; FIG. 17B is a circuit diagram of the X-address generating section that produces X-addresses X0 (A0R) to X9 (A9R); and FIG. 17C is a circuit diagram of the X-address generating section that produces X-addresses X10 (A10R) to X11 (A11R).

As shown in FIG. 17A, the address generating section 216, which is supplied with an address input Ain, produces an address Aj and its inverted address BAj from the address input Ain on the basis of the row address accept signal RACP. In this embodiment, 12 address generating sections 216 of FIG. 17A are used. These sections 216.sub.0 to 216.sub.11 generate addresses A0 (BA0) to A11 (BA11), respectively. In FIG. 17A, BRHLD indicates a row address hold signal (B means the inversion of signal level), BRLTC a row address latch signal (B means the inversion of signal level), and VRAD a reference potential.

The addresses A0 (BA0) to A11 (BA11) produced at the address generating sections 216.sub.0 to 216.sub.11 are supplied to the X-address generating sections 218.sub.0 to 218.sub.11 shown in FIGS. 17B and 17C. Based on the row address transfer signal BRTRS (B means the inversion signal level), the X-address generating sections 218.sub.0 to 218.sub.11 produce X-addresses X0 (A0R) to X11 (A11R) from addresses A0 (BA0) to A11 (BA11). Among the X-address generating sections 218.sub.0 to 218.sub.11, 218.sub.10 and 218.sub.11 have the circuit construction of FIG. 17C in order to cope with a change in the number of X-addresses due to the modification of refresh cycle. Specifically, each of X-address generating sections 218.sub.10 and 218.sub.11 contains NOR circuits 220 and 222, and X-address is supplied after passing through these NOR circuits 220 and 222. The first inputs of the NOR circuits 220 and 222 are supplied with signals C1 and C2, respectively. The X-address generating sections 218.sub.10 and 218.sub.11 supply X-addresses or not, depending on the signals C1 and C2. In this embodiment, the signals C1 and C2 are set as follows: in the generating section 218.sub.10 that produces X-address X10 (A10R), the first switching signal R1K is used as signals C1 and C2; and in the switching section 218.sub.11 that produces X-address X11 (A11R), the second switching signal R2K is used as signals C1 and C2. In FIGS. 17B and 17C, Cj and BC.sub.j indicate the counter outputs, and CTRS a count transfer signal.

With the X-address generating sections 218.sub.10 and 218.sub.11 of the aforesaid construction, when the switching signal R1K is in the H-level and the switching signal R2K is in the H-level (in the case of the 1-kcycle-refresh product), the generating sections 218.sub.10 and 218.sub.11 will not produce X-addresses X10 and X11. As explained in FIG. 16, the 1-kcycle-refresh product does not use X-addresses X10 and X11 (but uses Y-addresses Y8 and Y9). As a result, unnecessary X-addresses are not produced at the X-address buffer group 3, thereby reducing the power consumption, or preventing erroneous operations.

When the switching signal R1K is in the L-level and the switching signal R2K is in the H-level (in the case of the 2-kcycle-refresh product), the generating section 218.sub.10 will produce X-address X10, and the generating section 218.sub.11 will not produce X-addresses X11. Thus, as with the 1-kcycle-refresh product, unnecessary X-addresses are not produced at the X-address buffer group 3.

When the switching signal R1K is in the L-level and the switching signal R2K is in the L-level (in the case of the 4-kcycle-refresh product), the generating sections 218.sub.10 and 218.sub.11 will both produce X-addresses X10 and X11.

FIG. 18 is a circuit diagram of the Y-address buffer group 19 of FIG. 14. FIG. 18A is a circuit diagram of the Y-address generating section that produces Y-addresses YO (A0C) to Y7 (A7C) and FIG. 18B is a circuit diagram of the Y-address generating section that produces Y-addresses Y8 (A8C) to Y9 (A9C).

As shown in FIGS. 18A and 18B, the Y-address generating sections 224.sub.0 to 224.sub.9, which are supplied with address input Ain, produce Y-addresses YO (A0C) to Y9 (A9C) from address input Ain on the basis of a first column address latch signal CLTC and the second column address latch signal CLTD with a little delay behind the signal CLTC. Among the Y-address generating sections 224.sub.0 to 224.sub.9, 224.sub.8 and 224.sub.9 have the circuit arrangement of FIG. 18B in order to cope with a change in the number of Y-addresses due to the modification of refresh-cycle. Specifically, each of Y-address generating sections 224.sub.8 and 224.sub.9 contains NOR circuits 226 and 228, and Y-address is supplied after passing through the NOR circuits 226 and 228. The first inputs of the NOR circuits 226 and 228 are supplied with signals D1 and D2, respectively. The Y-address generating sections 224.sub.8 and 224.sub.9 supply Y-addresses or not, depending on signals D1 and D2. In this embodiment, signals D1 and D2 are set as follows: in the generating section 224.sub.8 that produces Y-address Y8 (A8C), the switching signal BR2K, the inversion in signal level of the second switching signal R2K, is used as signals D1 and D2; and in the generating section 224.sub.9 that produces Y-address Y9 (A9C), the switching signal BR1K, the inversion in signal level of the first switching signal R1K, is used as signals D1 and D2.

With the Y-address generating sections 224.sub.8 and 224.sub.9 of the aforesaid construction, when the inverted switching signal BR1K is in the L-level and the inverted switching signal BR2K is in the L-level (in the case of the 1-kcycle-refresh product), the generating sections 224.sub.8 and 224.sub.9 produce Y-addresses Y8 and Y9, respectively.

When the inverted switching signal BR1K is in the H-level and the inverted switching signal BR2K is in the L-level (in the case of the 2-kcycle-refresh product), the generating section 224.sub.8 will produce Y-address Y8, and the generating section 224.sub.9 will not produce Y-address Y9.

When the inverted switching signal BR1K is in the H-level and the inverted switching signal BR2K is in the H-level (in the case of the 4-kcycle-refresh product), none of the generating sections 224.sub.8 and 224.sub.9 produce Y-address Y8 or Y9.

FIG. 19 is a circuit diagram of the counter circuit group 37 of FIG. 14. FIG. 19A is a circuit diagram of a counter that produces counter outputs C.sub.0 to C.sub.9 ; FIG. 19B is a circuit diagram of a counter that produces counter output C.sub.10 ; and FIG. 19C is a circuit diagram of a counter that produces counter output C.sub.11.

As shown in FIG. 19A, the counter 230.sub.0, which is supplied with counter transfer signal CTRS (BCTRS), supplies counter output C.sub.0 (BC.sub.0) based on the signal CTRS (BCTRS). The counter 230.sub.1, which is supplied with counter output C.sub.0 (BC.sub.0), supplies counter output C, (BC.sub.1) based on counter output C.sub.0 (BC.sub.0). Subsequent counters are connected the same way, and the counter 230.sub.1 is supplied with counter output C.sub.9 (BC.sub.9) as shown in FIGS. 19B and 19C. The counter 230.sub.10 supplies counter output C.sub.10 (BC.sub.10) based on counter output C.sub.9 (BC.sub.9). The counter 230.sub.11, which is supplied with counter output C.sub.10 (BC.sub.10), supplies counter output C.sub.11 (BC.sub.11) based on counter output C.sub.10 (BC.sub.10). Among counters 230.sub.0 to 230.sub.11, 231.sub.10 and 230.sub.11 have the circuit arrangement of FIGS. 19B and 19C in order to cope with a change in the number of X-addresses due to the modification of refresh cycle. Specifically, the counter 230.sub.10 contains a clocked inverter 232.sub.10 that is turned on or off based on the switching signal BR1K, the inversion in signal level of the switching signal R1K. The counter 230.sub.11 contains a clocked inverter 232.sub.11 that is turned on or off based on the switching signal R4K. Thus, depending on the switching signals R1K and R4K, the counters 232.sub.10 and 232.sub.11 supply the counter signal or not.

With the counters 232.sub.10 and 232.sub.11 of the aforesaid construction, when the switching signal R1K is in the H-level and the switching signal R4K is in the L-level (in the case of the 1-kcycle-refresh product), the counters 232.sub.10 and 232.sub.11 will not produce counter outputs C.sub.10 and C.sub.11.

When the switching signal R1K is in the L-level and the switching signal R4K is in the L-level (in the case of the 2-kcycle-refresh product), the counter 232.sub.10 will produce counter output C.sub.10, and the counter 232.sub.11 will not produce counter output C.sub.11.

When the switching signal R1K is in the L-level and the switching signal R4K is in the H-level (in the case of the 4-kcycle-refresh product), the counters 232.sub.10 and 232.sub.11 will produce counter outputs C.sub.10 and C.sub.11.

FIG. 20 is a circuit diagram of the word-line boosting section 41 of FIG. 14.

As shown in FIG. 20, the word-line boosting section 41 is supplied with the first and second switching signals R1K and R2K. The boosting section 41 supplies the boosting capacitance WKM based on the signal .phi.WL commanding the boosting start. This section 41 contains a NOR gate 234 and NAND gates 236 and 238. The NOR gate 234 has a first input supplied with the switching signal R1K, and a second input with the switching signal R2K. The NAND gate 236 has a first input supplied with the signal R1K, and a second input with the signal .phi.WL. The NAND gate 238 has a first input supplied with the inversion in signal level of the output of NOR gate 234, and a second input with the signal .phi.WL.

With the word-line boosting section 41 of the above-described construction, when the switching signal R1K is in the H-level and the switching signal R2K is in the H-level (in the case of the 1-kcycle-refresh product), bringing signal .phi.WL into the H-level allows one electrode of a first capacitor 240.sub.1 to go to the H-level. Similarly, one electrode of each of a second and third capacitors 240.sub.2 and 240.sub.3 also goes to the H-level. Therefore, in the 1-kcycle-refresh product, the boosting capacitance potential WKM is produced by using capacitors 240.sub.1 to 240.sub.3.

When the switching signal R1K is in the L-level and the switching signal R2K is in the H-level (in the case of the 2-kcycle-refresh product), bringing signal .phi.WL into the H-level allows one electrode of the first capacitor 240.sub.1 to go to the L-level, and one electrode of each of the second and third capacitors 240.sub.2 and 240.sub.3 to go to the H-level. Therefore, in the 2-kcycle-refresh product, the boosting capacitance WKM is produced by using capacitors 240.sub.2 to 240.sub.3.

When the switching signal R1K is in the L-level and the switching signal R2K is in the L-level (in the case of the 4-kcycle-refresh product), bringing signal .phi.WL into the H-level allows one electrode of each of the first and second capacitors 240.sub.1 and 240.sub.2 to go to the L-level, and one electrode of the third capacitor 240.sub.3 alone to go to the H-level. Therefore, in the 4-kcycle-refresh product, the boosting capacitance WKM is produced by using capacitor 240.sub.3 only.

FIG. 23 shows how addresses are allocated. FIG. 23A shows the address allocation for the 1-kcycle-refresh product (mode); FIG. 23B for the 2-kcycle-refresh product (mode); and FIG. 23C for the 4-kcycle-refresh product (mode).

FIG. 24 is a block diagram showing the construction of the I/O sense amplifier groups 13.sub.0 to 13.sub.3 of FIG. 1.

As shown in FIG. 24, the I/O sense amplifier groups 13.sub.0 to 13.sub.3 contain sense circuits S and select circuits 300.sub.00 to 300.sub.31 for selecting sense circuits S. The sense circuits S are supplied with outputs I/O.sub.00 to I/O.sub.31 from the sense amplifiers 11.sub.0 to 11.sub.3. The select circuits 300.sub.00 to 300.sub.31, which are supplied with signals E and F, produce signals for selecting a desired sense circuit S based on signals E and F. Signal E is the output from the Y2-decoder 23, and signal F is the output of the I/O sense amplifier control circuit 31. The output of the sense circuit S selected by the select circuits 300.sub.00 to 300.sub.31 is, for example, output data D.sub.OUT.

The I/O sense amplifier groups 13.sub.0 to 13.sub.3 of the above-described construction has the advantages of decreasing the number of data output lines 302 and simplifying the circuit arrangement of the data input/output system.

FIG. 25 is a block diagram showing another construction of the I/O sense amplifier groups 13.sub.0 to 13.sub.3 of FIG. 1.

As shown in FIG. 25, the I/O sense amplifier groups 13.sub.0 to 13.sub.3 contain sense circuits S and select circuits 300.sub.0 to 300.sub.3 for selecting the I/O sense amplifier groups 13.sub.0 to 13.sub.3. The sense circuits S are supplied with outputs I/O.sub.00 to I/O.sub.31 from the sense amplifiers 11.sub.0 to 11.sub.3. The select circuits 300.sub.0 to 300.sub.3, which are supplied with signal F, produce signals for selecting a desired sense amplifier group 13.sub.0 to 13.sub.3 based on signal F. Signal F is the output of the I/O sense amplifier control circuit 31. The output signal from the sense amplifier group S selected by the select circuits 300.sub.0 to 300.sub.3 is supplied to a multiplexer circuit 304, which selects a desired sense circuit S based on signal E, for example. The signal E is the output of the Y2 decoder 23. The output of the sense circuit S selected by the multiplexer circuit 304 is, for example, output data D.sub.OUT.

The I/O sense amplifier groups 13.sub.0 to 13.sub.3 of the above-described construction has the advantage of simplifying the circuit arrangement of the I/O sense amplifier groups 13.sub.0 to 13.sub.3.

FIG. 26 is a flowchart of a chip selecting method according to the present invention.

This flowchart is used with a device that determines the product specification according to the bonding option shown in FIG. 2.

As shown in FIG. 26, at step 1, a pretreatment wafer process is performed to form DRAM chips (integrated circuit chips) in the wafer. After the DRAM chips have been formed, at step 2, a chip screening test is made to see whether the formed DRAM chips are acceptable or not. After this, a pause test (a data retention test) is carried out to determine how long the memory cell in the DRAM chips can retain the data. At step 3, redundancy fuse-cut is performed to save the chips judged to be unacceptable at the step 2 chip screening test, to some extent (redundancy techniques). At step 4, the wafer undergoes dicing, which divides the wafer into a plurality of DRAM chips. At step 5, the chips are assembled. In this process, each chip is mounted on a bed and the chip's pad is bonded to a lead. At this time, based on the result of the step 2 pause test, bonding is done to select a refresh-cycle mode. This process is done depending on whether a wire is bonded to the bonding pad P of the receiving section 27 of FIG. 2. This bonding determines, for example, the 2-kcycle-refresh product (mode) or the 4-kcycle-refresh product (mode) semipermanently. Then, the packaging process is carried out to form the final product. After this, at step 6, a final test is performed, and the products that have passed this test are put on the market.

FIG. 27 is another flowchart of a chip screening select method according to the present invention.

This flowchart is used with a device that determines the product specification according to the fuse option shown in FIG. 3.

As shown in FIG. 27, at step 3, redundancy fuse-cutting is done. In this step, refresh-cycle select fuse cutting is also done. In this process, the fuse F of the receiving section 27 of FIG. 3 is blown or not. As with the method of FIG. 26, this fuse cutting determines, for example, the 2-kcycle-refresh product (mode) or the 4-kcycle-refresh product (mode) semipermanently.

Since the chip select method determines the 2-kcycle-refresh product (mode) or the 4-kcycle-refresh product (mode) on the basis of the result of the pause test, even if, for example, chips with memory cells whose pause time is shorter than the design pause time due to variations in the processes, they may be saved as the 4-kcycle-refresh product (mode), thereby improving the product yield.

Even in the course of manufacturing, it is easy to change the product specification from the 2-kcycle-refresh product (mode) to the 4-kcycle one or vice versa, providing flexibility in manufacturing products.

FIG. 28 shows the contents of step 2 in FIGS. 26 and 27 in detail.

As shown in FIG. 28, the tests at step 2 are broadly divided into two tests: a chip screening test and a pause test. Of these tests, the chip screening test is further divided into subtests: for example, an operating current test, a typical voltage test, a cell to cell interference test, and others. Each test has its own optimum refresh-cycle. Therefore, setting the optimum refresh-cycle before each test makes it possible to shorten the test time and improve the select capability, thereby improving the chip select test efficiency.

For example, the operating current test in test item TEST A is made with a 2-kcycle-refresh. With the operating current test with a 2-kcycle-refresh, the chip select conditions can be made more severe than those with a 4-kcycle-refresh, making it possible to select only chips with very high reliability.

The typical voltage test in test item TEST B is carried out with a 4-kcycle-refresh. In the typical voltage test with a 4-kcycle-refresh, the short-circuit of word lines (for example, adjacent word lines) that is unacceptable in the 2-kcycle-refresh product is acceptable in the 4-kcycle-refresh product, thereby increasing the number of acceptable products. When 2-kcycle-refresh products are to be obtained from the lot subjected to this test, however, there is a possibility that unacceptable products may also be included in them. To avoid this problem, the typical voltage test with 2-kcycle-refresh should be made. When only 4-kcycle-refresh products are obtained, the typical voltage test with 2-kcycle-refresh may not be performed. In this way, the test may be made with 2-kcycle-refresh or 4-kcycle-refresh as required.

The cell-to-cell interference test in test item TEST C is made with the 2-kcycle-refresh product. The cell-to-cell interference test with a 2-kcycle-refresh allows current to flow to all memory cells in a shorter time than that with 4-kcycle-refresh, thereby shortening the test time.

For the tests not shown in FIG. 28, the respective optimum refresh-cycles are set similarly.

FIG. 29 is a sectional view of the pad P of FIG. 2.

Setting the optimum refresh-cycle for each test can be achieved by simply bringing the probe 28 of the wafer prober into contact with the bonding pad P as shown in FIG. 29, and applying a voltage to the receiving section 27 or not.

The present invention is not limited to the above embodiments, and may be practiced or embodied in still other ways without departing from the spirit or essential character thereof. For example, in the foregoing embodiments, the decision signal SDS to determine the product specification is supplied to the receiving section 27 by means of wire bonding or the cutting of fuse F. Instead of the fuse F, a nonvolatile memory cell may be used to supply the decision signal SDS depending on whether the cell turns on or not.

Also, the package may have an additional pin, to which the signal SDS is supplied, so that the user can select one of the two refresh-cycle modes by supplying the signal SDS to the additional pin, and the other refresh-cycle mode by not supplying the signal SDS to the additional pin. Further, the package may have two or more additional pins, to which the signals VR1K, VR2K are supplied, so that the user can select any desired one of three or more refresh-cycle modes by supplying the signal SDS to one or more of the additional pins.

Additional advantages and modifications will readily occur to those skilled in the art. Therefore, the invention in its broader aspects is not limited to the specific details, representative devices, and illustrated examples shown and described herein. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents.