System for Driving Bipolar Ram

1. Field of the Invention

The present invention relates generally to a bipolar memory system and, more particularly, to a system for driving a bipolar RAM (Random Access Memory).

2. Description of the Prior Art

In general, a bipolar RAM comprises a cell array composed of memory cells each of which includes a flip-flop formed by two transistors cross-coupled as a fundamental unit. Each of the memory cells is connected to a word line (W), a hold line (H) and a pair of bit lines (B, B) by which the memory cells can be selected. In this case, selection of one word line, i.e. raising the potential thereof, is effected by a work decoder composed of row address buffers and word drivers.

One prior art system for driving a bipolar RAM is characterised in that the difference in potential between the selected word line and the non-selected word lines is substantially the same during the read mode and the write mode. However, usually, since the range of potential of the write operation is larger than that of the read operation, the above-mentioned difference in potential is set at the margin of operation of the write operation. As a result, during the read mode, when the state of one word line is changed from the non-selected to the selected mode, in other words, when the potential of one word line is changed from low to high, the time required to charge the word line with stray capacity is long, so that the reading speed of the memory cell is low.

In order to improve the reading speed performance, a system is known in the art for driving a bipolar RAM wherein the state of the selected word line is changed from the read mode to the write mode, whereby the voltage of the selected word line is forced to be higher (Ref. Japanese Public Disclosure No. 50-139635). Therefore, the difference in potential between the selected word line and the non-selected word lines during the read mode is set to be smaller than that during the write mode. As a result, the reading speed of the memory cell increases. However, in this system, as will be explained hereinafter, when the state of the selected word line is changed from the write mode to the read mode, the data written into the selected memory cell belonging to the selected word line may be destroyed.

It is a principal object of the present invention to provide a system for driving a bipolar RAM with high-speed and stable reading operation which incurs no data destruction in the read mode after the write mode.

According to the present invention, there is provided a system for driving a bipolar RAM, whose memory cells are selected by changing the potentials of word lines, comprising: a first means for changing the potentials of the word lines responsive to the selected or non-selected state thereof, the potential of the one selected word line being higher than that of the non-selected word lines; and a second means for changing the potentials of the non-selected word lines responsive to the read or write mode thereof, the potentials of said non-selected word lines being lower during the write mode than during the read mode, whereby the difference in potential between the selected word line and the non-selected word lines is smaller during the read mode than during the write mode. By using this system, the data written into the memory cells cannot be destroyed during the read mode after the write mode. In addition. during the read mode, the difference in potential between the selected word line and the non-selected word lines is set at the lower limit of the margin of operation for the read operation, so that the reading speed of the memory cells becomes high.

The present invention will be more clearly understood from the following description with reference to the accompanying drawings, wherein:

FIG. 1 is a block diagram illustrating the memory cell array of a general bipolar RAM whose memory cells are of a Schottky clamped type;

FIG. 2 is the main portion of FIG. 1;

FIG. 3 is a timing diagram showing the signals appearing in the circuit of FIG. 2, by using a driving system of the prior art;

FIG. 4 is a timing diagram showing the signals appearing in the circuit of FIG. 2, by using another driving system of the prior art;

FIG. 5 is a timing diagram showing the signals appearing in the circuit of FIG. 2, by using the driving system according to the present invention;

FIG. 6 is a block diagram illustrating the memory cell array of another bipolar RAM whose memory cells are of an IIL (Integrated Injection Logic) type;

FIG. 7 is a circuit diagram illustrating a general decoder/driver circuit;

FIGS. 8A and 8B are two kinds of circuit diagrams of the address buffer BF.sub.i of the decoder/driver circuit of FIG. 7;

FIGS. 9A, 9B and 9C are three kinds of circuit diagrams of the driver portion of the decoder/driver circuit of FIG. 7;

FIG. 10 is a circuit diagram illustrating another general decoder/driver circuit;

FIG. 11 is a circuit diagram of the address buffers BF.sub.0 ' and BF.sub.1 ' of the decoder/driver circuit of FIG. 10;

FIG. 12 is a circuit diagram of the driver portion of the decoder/driver circuit of FIG. 10;

FIG. 13 is a circuit diagram illustrating a first embodiment of the decoder/driver circuit of the present invention;

FIG. 14 is a circuit diagram of the current switching circuits CS1 of FIG. 13;

FIGS. 15 and 16 are circuit diagrams illustrating second and third embodiments of the decoder/driver circuit of the present invention, respectively;

FIG. 17 is a circuit diagram of the current switching circuit CS2 of FIG. 16; and

FIGS. 18, 19 and 20 are circuit diagrams illustrating fourth, fifth and sixth embodiments of the decoder/driver circuit of the present invention.

In FIG. 1, a memory array composed of memory cells C.sub.ij (i=0, 1, . . ., n-1; j=0, 1, . . ., m-1) whose matrix is n rows by m columns is shown. The memory cells C.sub.ij which are of a nonsaturating type (Schottky clamped type) comprise two cross-coupled multi-emitter transistors, the transistor elements of which are indicated by Q.sub.C1 and Q.sub.C2 whose emitters are connected to bit lines, and Q.sub.H1 and Q.sub.H2 whose emitters are connected to a hold line, two Schottky barrier diodes SBD.sub.1 and SBD.sub.2, and two load resistors R.sub.1 and R.sub.2, whereby a flip-flop is formed. The potentials at the collectors of the transistor elements Q.sub.C1, Q.sub.H1, Q.sub.C2, and Q.sub.H2 are clamped by the Schottky barrier diodes SBD.sub.1 and SBD.sub.2, so that the transistor elements do not enter into the saturation state. For example, the memory cell C.sub.00 is located between a word line W.sub.D and a hold line H.sub.D through which a constant current I.sub.H flows and between a pair of bit lines B.sub.0 and B.sub.0. The word line W.sub.D is selected by a word driver WD.sub.0 and the bit line pair B.sub.0 and B.sub.0 are selected by a bit driver BD.sub.0 (not shown) so that the memory cell C.sub.00 can be selected. However, it is also possible to select the memory cell C.sub.00, even if only the word line W.sub.0 and all the bit line pairs are selected. The sense amplifier SA detects the data stored in the memory cells during the read mode. The write amplifier WA produces output voltages V.sub.D and V.sub.D whose values are different when the potential of a write-enable signal WE is low, which means the write mode, and whose values are the same when the potential of the write-enable signal WE is high, which means the read mode.

The operation of the memory cell array of FIG. 1 will be explained by using FIG. 2, which is the main portion of FIG. 1, and FIG. 3 which is a timing diagram showing the signals appearing in the circuit of FIG. 2. The access operation with regard to the memory cell C.sub.00 (FIG. 1) is shown by the progression of the following steps, as illustrated in FIG. 3.

Step 1: Non-selected state;

Step 2: The read mode in the selected state;

Step 3: The write mode in the selected state;

Step 4: The read mode in the selected state; and

Step 5: Non-selected state.

In step 1, it is assumed that the transistor elements Q.sub.H1 and Q.sub.H2 of the memory cell C.sub.00 which form a flip-flop are conductive and non-conductive respectively. In addition, the transistor elements Q.sub.C1 and Q.sub.C2 of the memory cell C.sub.00 are non-conductive, since the cell C.sub.00 is non-selected. Further, the voltages V.sub.D and V.sub.D of the write amplifier WA(FIG. 1) are the same, which means the read mode. The collector potential V.sub.BS is represented as follows.

where R is the resistance of the resistor R.sub.2 and I.sub.B1 is the base current of the transistor element Q.sub.H1. In this case, the value R.times.I.sub.B1 is very small so that the Schottky barrier diode SBD.sub.2 cuts off. Contrary to this, the collector voltage V.sub.CS is represented as follows.

where V.sub.BE is the forward voltage of the Schottky barrier diode SBD.sub.1. In this case, the collector current I.sub.C1 is relatively large so that the Schottky barrier diode SBD.sub.1 is conductive.

In step 2, the voltage V.sub.W0 of the word line W.sub.0 is increased so that the collector voltages V.sub.BS and V.sub.CS are also increased, while the voltage V.sub.W1 of the word line W.sub.1 is decreased, as illustrated in FIG. 3. The voltages V.sub.D and V.sub.D, which are the same, are preset to be between the voltages V.sub.BS and V.sub.CS. Preferably, the voltages V.sub.D and V.sub.D equal (V.sub.BS +V.sub.CS)/2. In this case, since the transistor element Q.sub.C1, to whose base is applied the voltage V.sub.BS, and a transistor Q.sub.0, to whose base is applied the voltage V.sub.D which is smaller than the voltage V.sub.BS form a current switch, the transistor element Q.sub.C1 conducts and the transistor Q.sub.0 remains non-conductive. Similarly, since the transistor elements Q.sub.C2 and the transistor Q.sub.0 ' form a current switch, the transistor element Q.sub.C1 remains non-conductive and the transistor Q.sub.0 ' conducts. As a result, currents indicated by arrows X1 and X2 flow through data bit lines DB and DB, respectively. Therefore, the currents I.sub.D and I.sub.D supplied to the sense amplifier SA are represented as follows.

In this case, the output data D.sub.OUT of the sense amplifier SA is, for example, the logic "0".

In step 3, the voltage V.sub.D is decreased which means the write mode for writing the logic "1". When the voltage V.sub.D equals the voltage V.sub.CS at time t.sub.1, the emitter currents of the transistor element Q.sub.C2 and the transistor Q.sub.0 ' becomes equal (=1/2 I.sub.B). Furthermore, when the voltage V.sub.D becomes lower than the voltage V.sub.CS, the emitter current of the transistor element Q.sub.C2 becomes equal to I.sub.B and the emitter current of the transistor Q.sub.0 ' becomes zero. In this case, the difference in potential between the voltages V.sub.CS and V.sub.D is, for example, higher than 200 mV. As a result, the voltages V.sub.BS and V.sub.CS become low and high, respectively, so that the state of the flip-flop is reversed. In this write mode, the voltage V.sub.D is increased in order to spend up the write operation. However, it should be noted that the voltage V.sub.D can remain at the same level as the read mode. In this case, the write speed becomes lower.

In step 4, the levels of the voltages V.sub.D and V.sub.D are the same. In this case, currents indicated by arrows Y1 and Y2 flow through the data bit lines DB and DB, respectively. Therefore, the currents I.sub.D and I.sub.D supplied to the sense amplifier SA are represented as follows.

In this case, the output data D.sub.OUT of the sense amplifier SA is the logic "1".

In step 5, the voltage V.sub.W0 of the word line W.sub.0 is decreased so that the collector voltages V.sub.BS and V.sub.CS are also decreased. However, the decrease of the voltages V.sub.BS and V.sub.CS takes a longer time than the decrease of the voltage V.sub.W0. Contrary to this, the voltage V.sub.W1 of the word line W.sub.1 is increased.

In steps 2, 3 and 4 wherein the memory cell C.sub.00 is selected and the memory cell C.sub.01 is non-selected, both of the voltages V.sub.BN and V.sub.CN of the memory cell C.sub.01 should always be lower than the voltage V.sub.D and V.sub.D so that the transistor elements Q.sub.C1 and Q.sub.C2 thereof are always nonconductive. Therefore, access into the memory cell C.sub.01 is never effected.

In the above-mentioned access operation, there are some operating requirements. One such requirement is that the voltage V.sub.D and the voltage V.sub.W1 during the write mode in the selected state (Step 3) should satisfy the following inequality.

In other words, the voltages V.sub.D and V.sub.D should be higher than the voltage of the non-selected word line. Otherwise, if the transistor element Q.sub.H2 of the non-selected memory cell C.sub.01 is conductive, the transistor element Q.sub.C2 conducts so that current from the non-selected memory cell C.sub.01 flows through the bit line B.sub.0. As a result, it takes a long time for the transistor element Q.sub.C2 of the selected memory cell C.sub.00 to conduct, which means that the write speed for the selected memory cell is low. At worst, the write operation may not occur. It should be noted that the difference .DELTA.V.sub.1 in potential between the volgates V.sub.BS and V.sub.CS is about 400 or 500 mV and the difference .DELTA.V.sub.2 in potential between the voltages V.sub.CS and V.sub.D is about 200 mV in the non-saturating type memory cells using Schottky barrier diodes (FIG. 1). In addition, the difference .DELTA.V.sub.1 in potential is about 700 mV in the saturating type memory cells using p-n junction diodes.

Another such condition is that the voltage V.sub.W1 and the voltage V.sub.D (=V.sub.D) during the read mode in the selected state (Steps 2 and 4) should satisfy the following inequality.

In other words, the voltage of the non-selected word line should be lower than the voltage V.sub.D (=V.sub.D). Otherwise, in the case of step 2, the current I.sub.B flowing through the data bit line DB comes from both the transistor Q.sub.0 ' and the transistor element Q.sub.C2 of the non-selected read mode, and the time (which corresponds to .DELTA.t in FIG. 3) required to charge the word line with its stray capacitance is long, so that the reading speed of the memory cell is slow.

FIG. 4 is a timing diagram showing signals appearing in the circuit of FIG. 2, by using another driving system of the prior art (Ref. Japanese public Disclosure No. 50-139635). In FIG. 4, when the state of the selected word line W.sub.0 is changed from the read mode to the write mode, the voltage V.sub.W0 of the selected word line W.sub.0 is changed higher, as indicated by an arrow A. Therefore, during the read mode, the difference .DELTA.W' in potential between the selected word line W.sub.0 and the non-selected word line W.sub.1 is smaller than the difference .DELTA.W(FIG. 3) in potential. As a result, the time (which corresponds to .DELTA.t') required to charge the selected word line with its stray capacitance is smaller than the time .DELTA.t (FIG. 3), which means that the reading speed of the operation as illustrated in FIG. 4 is faster than that as illustrated in FIG. 3. However, as illustrated in FIG. 4, when the state of the selected word line W.sub.0 is changed from the write mode to the read mode, the voltages V.sub.BS and V.sub.CS decrease more slowly than the voltages V.sub.D and V.sub.D. As a result, a phenomenon occurs where the voltage V.sub.D is lower than the voltage V.sub.BS (as indicated in FIG. 4 by area B) so that the transistor element Q.sub.C1 conducts and, accordingly, the transistor element Q.sub.H1 also conducts. This means that the data stored in the selected memory cell C.sub.00 is destroyed, since the transistor element Q.sub.H2 was already conductive.

FIG. 5 is a timing diagram showing the signals appearing in the circuit of FIG. 2, by using the driving system according to the present invention. In FIG. 5, when the state of the selected word line W.sub.0 is changed from the read mode to the write mode, the voltage V.sub.W1 of the non-selected word line W.sub.1 is changed to be lower, as indicated by an arrow C. Therefore, in the same manner as in FIG. 4, during the read mode, the difference .DELTA.W' in potential between the selected word line W.sub.0 and the non-selected word line W.sub.1 is small, so that the time which corresponds to .DELTA.t' required to charge the selected word line and its stray capacitance is small. Therefore, the reading speed of the operation, as illustrated in FIG. 5, is also high. Further, after the state of the selected memory cell C.sub.00 is changed from the write mode to the read mode, the voltage V.sub.D is never lower than the voltage V.sub.BS, so that the data stored in the selected memory cell C.sub.00 will not be destroyed.

FIG. 6 is a block diagram illustrating the memory cell array of another general bipolar RAM, whose memory cells are of an IIL (Integrated Injection Logic) type. In the read mode of the memory cell array of FIG. 6, for example, the difference in potential between bit lines B.sub.0 and B.sub.0 is amplified by a differential amplifier formed by two transistors Q.sub.S1 and Q.sub.S2, so that the amplified signals are supplied to a sense amplifier SA. In the write mode of the memory cell array of the FIG. 6, one of two transistors Q.sub.B0 and Q.sub.B1 is conductive and the other is non-conductive. Therefore, the conditions represented by the above-mentioned inequalities (7) and (8) are unnecessary. However, it is preferable, during the read mode, for the difference in potential between the selected word line and the non-selected word lines to be small, which permits a fast reading operation. Therefore, the driving system whose operation is illustrated in FIG. 5 is applied to the memory cell array of FIG. 6.

FIG. 7 is a circuit diagram illustrating a general decoder/driver circuit. In FIG. 7, it is assumed that there are sixteen word lines W.sub.0, W.sub.1, . . . , W.sub.15 so that four address signals A.sub.0, A.sub.1, A.sub.2, and A.sub.3 (2.sup.4 =16) are necessary. The address signals A.sub.0, A.sub.1, A.sub.2, and A.sub.3 are applied to address buffers BF.sub.0, BF.sub.1, BF.sub.2, and BF.sub.3, respectively, so that the address buffers produce the signals A.sub.0, A.sub.1, A.sub.2, and A.sub.3 and their inverted signals A.sub.0, A.sub.1, A.sub.2 and A.sub.3 to bus lines L.sub.0, L.sub.1, . . . , L.sub.7. Each of sixteen word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 have four inputs, which are connected to one of the lines L.sub.0 and L.sub.1, one of the lines L.sub.2 and L.sub.3, one of the lines L.sub.4 and L.sub.5, and one of the lines L.sub.6 and L.sub.7, respectively. Therefore, only one of the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 is switched on, and the others are switched off. In other words, only one of the word lines W.sub.0, W.sub.1, . . . , W.sub.15 is selected and the other word lines are non-selected.

FIGS. 8A and 8B are two kinds of circuit diagrams of the address buffer BFi of the decoder/driver circuit of FIG. 7. In FIG. 8A, two transistors Q.sub.1 and Q.sub.2 form a current switch. When the voltage of the address signal Ai(IN) is lower than a reference voltage V.sub.R1, the transistors Q.sub.1 and Q.sub.2 are conductive and non-conductive, respectively, so that the base voltages of the transistors Q.sub.3 and Q.sub.4 are low and high, respectively. As a result, the voltages of signals Ai(OUT) and Ai(OUT) are low and high, respectively. Contrary to this, when the voltage of the address signal Ai(IN) is higher than the reference voltage V.sub.R1, the voltages of signals Ai(OUT) and Ai(OUT) are high and low, respectively.

In FIG. 8B, two transistors Q.sub.5 and Q.sub.6 form a current switch. Therefore, when the voltage of the address signal Ai(IN) is lower than a reference voltage V.sub.R1, the transistors Q.sub.5 and Q.sub.6 are conductive and non-conductive, respectively, so that current signals Ai(OUT) and Ai(OUT) are in an off-state and on-state, respectively. Contrary to this, when the voltage of the address signal Ai(IN) is higher than the reference voltage V.sub.R1, the current signals Ai(OUT) and Ai(OUT) are in an on-state and off-state, respectively.

FIGS. 9A, 9B and 9C are three kinds of circuit diagrams of the driver portion of the decoder/driver circuit of FIG. 7. In FIG. 9A, transistors Q.sub.7, Q.sub.8, Q.sub.9, Q.sub.10, and Q.sub.11 form a current switch. Therefore, when all the base voltages of the transistors Q.sub.8, Q.sub.9, Q.sub.10, and Q.sub.11 are lower than a reference voltage V.sub.R2 of the transistor Q.sub.7, all the transistors Q.sub.8, Q.sub.9, Q.sub.10, and Q.sub.11 are non-conductive, so that the base voltage V.sub.X of a transistor Q.sub.X is high (.apprxeq.V.sub.CC) and, accordingly, the voltage V.sub.Wi of the word line W.sub.i is high. This means that the word line W.sub.i is selected. Contrary to this, when at least one of the base voltages of the transistors Q.sub.8, Q.sub.9, Q.sub.10, and Q.sub.11 are higher than the reference voltage V.sub.R2, a current flows through a resistor R.sub.x so that the voltage V.sub.X is low and, accordingly, the voltage V.sub.Wi of the word line W.sub.i is low. This means that the word line W.sub.i is non-selected. In FIG. 9B, when all diodes D.sub.1, D.sub.2, D.sub.3, and D.sub.4 are non-conductive, the voltage V.sub.X is high (.apprxeq.V.sub.CC) and, accordingly, the voltage V.sub.Wi of the word line W.sub.i is high (selected state). Contrary to this, when at least one of the diodes is conductive, the voltage V.sub.X is low and, accordingly, the voltage V.sub.Wi of the word line W.sub.i is low (non-selected state). In FIG. 9C, a multi-emitter transistor is shown, whose elements are indicated by Q.sub.12, Q.sub.13, Q.sub.14, and Q.sub.15. The transistor elements Q.sub.12, Q.sub.13, Q.sub.14, and Q.sub.15 correspond to the diodes D.sub.1, D.sub.2, D.sub.3, and D.sub.4, respectively. Therefore, the operation of the word driver of FIG. 9C is similar to that of FIG. 9B.

Returning to FIG. 7, in a case where the address buffers BF.sub.0, BF.sub.1, BF.sub.2, and BF.sub.3 are those of FIG. 8A, the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 should be those of FIG. 9A. On the other hand, in a case where the address buffers BF.sub.0, BF.sub.1, BF.sub.2, and BF.sub.3 are those of FIG. 8B, the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 should be those of FIG. 9B or 9C.

FIG. 10 is a circuit diagram illustrating another general decoder/driver circuit. In FIG. 10, two address signals A.sub.0 and A.sub.1 are converted into four signals A.sub.0 A.sub.1, A.sub.0 A.sub.1, A.sub.0 A.sub.1, and A.sub.0 A.sub.1 by address buffers BF.sub.0 ' and BF.sub.1 ', while two address signals A.sub.2 and A.sub.3 are converted into four signals A.sub.2 A.sub.3, A.sub.2 A.sub.3, A.sub.2 A.sub.3 and A.sub.2 A.sub.3 by address buffers BF.sub.2 ' and BF.sub.3 '. Here, an expression such as A.sub.0 A.sub.1 denotes a logical sum between A.sub.0 and A.sub.1. Each of the sixteen word drivers WD.sub.0 ', WD.sub.1 ', . . . , WD.sub.15 ' has two inputs, which receive one of the signals A.sub.0 A.sub.1, A.sub.0 A.sub.1, A.sub.0 A.sub.1, and A.sub.0 A.sub.1 and one of the signals A.sub.2 A.sub.3, A.sub.2 A.sub.3 , A.sub.2 A.sub.3, and A.sub.2 A.sub.3, respectively. Therefore, only one of the word drivers WD.sub.0 ', WD.sub.1 ', . . . , WD.sub.15 ' is switched on and the others are switched off. In other words, only one of the word lines W.sub.0, W.sub.1, . . . , W.sub.15 is selected and the other word lines are non-selected.

FIG. 11 is a circuit diagram of the address buffers BF.sub.0 ' and BF.sub.1 ' of the decoder/driver circuit of FIG. 10. In FIG. 11, two transistors Q.sub.16 and Q.sub.17 form a current switch and two transistors Q.sub.18 and Q.sub.19 form a current switch. Therefore, two address signals A.sub.0 (IN) and A.sub.1 (IN) are converted into the signals A.sub.0 and A.sub.1 and their inverted signals A.sub.0 and A.sub.1 by the current switches. The signals A.sub.0, A.sub.1, A.sub.0, and A.sub.1 are supplied to four multi-emitter transistors Q.sub.20, Q.sub.21, Q.sub.22, and Q.sub.23 which form four current switches which produce signals A.sub.0 A.sub.1, A.sub.0 A.sub.1, A.sub.0 A.sub.1, and A.sub.0 A.sub.1.

FIG. 12 is a circuit diagram of the driver portion of the decoder/driver circuit of FIG. 10. The circuit of FIG. 12 is the same as that of FIG. 9A, except for the number of inputs. Therefore, the operation of the circuit of FIG. 12 is similar to that of FIG. 9A.

FIG. 13 is a circuit diagram illustrating a first embodiment of the decoder/circuit of the present invention. The elements in FIG. 13 which are identical to those of FIGS. 7 and 9A are denoted by the same reference numerals as used in FIGS. 7 and 9A. As understood from FIG. 13, each of the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 is the same as that of FIG. 9A, except that the constant current source denoted by Q.sub.XS is controlled by a current switching circuit CS1. The current switching circuit CS1 which receives a write enable signal WE produces a signal whose voltage is high or low depending upon whether it is for the write mode or the read mode. For example, in the non-selected word line W.sub.1, at least one of the transistors Q.sub.8, Q.sub.9, Q.sub.10, and Q.sub.11 are conductive so that the current flowing through the resistor R.sub.x (whose value is also represented by R.sub.x) is the same as the current I.sub.XS flowing the transistor Q.sub.XS. As a result, the voltage V.sub.X is low (=V.sub.CC -I.sub.XS.R.sub.x). In addition, the current I.sub.XS in the write mode is larger than that in the read mode and, accordingly, the voltage V.sub.X is lower in the write mode than in the read mode. Therefore, the voltage V.sub.W1 is lower in the write mode than in the read mode. Contrary to this, in the selected word line W.sub.0, all the transistors Q.sub.8, Q.sub.9, Q.sub.10, and Q.sub.11 are non-conductive so that the voltage V.sub.X is always high (.apprxeq.V.sub.CC) regardless of the write or read modes. Therefore, the decoder/driver circuit of FIG. 13 can be operated as illustrated in FIG. 5. The embodiment of FIG. 13 can be also used in FIG. 10.

FIG. 14 is a circuit diagram of the current switching circuit CS1 of FIG. 13. In FIG. 14, it is assumed that, in the write mode, the voltage of the write enable signal WE is lower than a reference voltage V.sub.R3, while in the read mode, the voltage of the signal WE is higher than the reference voltage V.sub.R3. Therefore, in the write mode, a transistor Q.sub.21 which forms a current switch with a transistor Q.sub.22 is conductive and the transistor Q.sub.22 is non-conductive. As a result, the base voltage of a transistor Q.sub.23 becomes high and, accordingly, the base voltage of a transistor Q.sub.25 becomes high. As a result, the current I.sub.XS is increased. The operation of the write mode will be explained in more detail. If it is assumed that each of the forward voltages of diodes D.sub.11 and D.sub.12 and each of the base-emitter voltages of transistors Q.sub.27 and Q.sub.28 are the same (=V.sub.BE), the collector current I.sub.C8 of the transistor Q.sub.28 is represented as follows. ##EQU1## where R.sub.12 represents the resistance value of a resistor R.sub.12 as well as the resistor R.sub.12 itself. If the emitter area of a transistor Q.sub.24 is the same as that of the transistor Q.sub.28, the collector current I.sub.C4 of the transistor Q.sub.24 equals I.sub.C8. Therefore, the voltage drop .DELTA.V of a resistor R.sub.14, whose value is also represented by R.sub.14, is represented as follows. ##EQU2## Here, if the value R.sub.14 equals the value R.sub.12, then

Since the transistor Q.sub.22 is non-conductive, the base voltage of the transistor Q.sub.23 is V.sub.CC -V.sub.BE, and the base voltage V.sub.B5 of the transistor Q.sub.25 is represented as follows. ##EQU3## Since the voltage V.sub.6 of a node connecting the transistor Q.sub.XS and a resistor R.sub.16 is obtained by subtracting the drop in voltage of the transistors Q.sub.25 and Q.sub.XS from the voltage V.sub.B5, the voltage V.sub.6 is represented by

Thus, the value V.sub.BE determines the constant current I.sub.XS in the write mode.

On the other hand, in the read mode, the transistors Q.sub.21 and Q.sub.22 are non-conductive and conductive, respectively. As a result, the base voltage of the transistor Q.sub.23 becomes lower by a voltage drop of I.sub.1 R.sub.11. Since this voltage drop cause a drop in the voltage V.sub.6 of magnitude I.sub.1.R.sub.11, the voltage V.sub.6 is represented as follows.

Thus, the value (V.sub.BE -I.sub.1.R.sub.11) determines the constant current I.sub.XS in the read mode.

FIG. 15 is a circuit diagram illustrating a second embodiment of the decoder/driver circuit of the present invention. The elements in FIG. 13 which are identical to those of FIGS. 7 and 8B are denoted by the same reference numerals as used in FIGS. 7 and 8B. As understood from FIG. 15, each of the address buffers BF.sub.0, BF.sub.1, BF.sub.2, and BF.sub.3 is the same as that of FIG. 8B, except that the constant current source denoted by Q.sub.I is controlled by the current switching circuit CS1 of FIG. 14. In this case, each of the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 is the same as that of FIG. 9B or 9C. The current I.sub.X is larger in the write mode than in the read mode. As a result, in the write mode, the output currents from the address buffers become large, which tends to cause an increase in current flowing through the resistor R.sub.X (FIG. 9B) or the resistors R.sub.X1 and R.sub.X2 (FIG. 9C) which belong to a non-selected word driver such as WD.sub.1. However, in the selected word driver such as W.sub.0, the voltage V.sub.X in FIG. 9B or 9C remains high (.apprxeq.V.sub.CC) regardless of the write or read mode. Therefore, the decoder/driver circuit of FIG. 15 can be operated as illustrated in FIG. 5.

FIG. 16 is a circuit diagram illustrating a third embodiment of the decoder/driver circuit of the present invention. The elements in FIG. 16 which are identical to those of FIGS. 7 and 8B are denoted by the same reference numerals as used in FIGS. 7 and 8B. As understood from FIG. 16, each of the address buffers BF.sub.0 BF.sub.1, BF.sub.2, and BF.sub.3 comprises two multi-emitter transistors Q.sub.5 ' and Q.sub.6 ', and a transistor Q.sub.7 '. With regard to a constant current source I.sub.X, the transistors Q.sub.5 ' and Q.sub.6 ' form a current switch, and with regard to a constant current source .DELTA.I.sub.X, the transistors Q.sub.5 ', Q.sub.6 ' and Q.sub.7 ' form a current switch. In addition, each of the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 is also the same as that of FIG. 9B or 9C. In each of the address buffers BF.sub.0, BF.sub.1, BF.sub.2, and BF.sub.3, the base voltage V.sub.RW of the transistor Q.sub.7 is controlled by a current switching circuit CS2. The current switching circuit CS2 which receives a write enable signal WE produces a signal V.sub.RW. In the read mode, the voltage V.sub.RW of the current switch circuit CS2 is higher than any of the reference voltages V.sub.R1 or the voltages of the address signals, so that the current .DELTA.I.sub.X flows through the transistor Q.sub.7 '. As a result, in the non-selected word driver WD.sub.1, the current flowing through the resistor R.sub.X (FIG. 9B) or the resistors R.sub.X1 and R.sub.X2 (FIG. 9C) remains I.sub.X. Contrary to this, in the write mode, the voltage V.sub.RW is lower than any of the reference voltages V.sub.R1 or the voltages of the address signals, so that the current (I.sub.X .DELTA.I.sub.X) flows only through one of the transistors Q.sub.5 ' and Q.sub.6 '. As a result, in the non-selected word driver WD.sub.1, the current flowing through the resistor R.sub.X (FIG. 9B) or the resistors R.sub.X1 and R.sub.X2 (FIG. 9C) becomes (I.sub.X +.DELTA.I.sub. X). However, in the selected word driver W.sub.0, the voltage V.sub.X in FIG. 9B or 9C remains high (.apprxeq.V.sub.CC) regardless of the write or read mode. Thus, the decoder/driver circuit of FIG. 16 can be also operated as illustrated in FIG. 5.

FIG. 17 is a circuit diagram of the current switching circuit CS2 of FIG. 16. In the write mode, since the voltage of the write enable signal WE is lower than the reference voltage V.sub.R3, transistors Q.sub.31 and Q.sub.32 which form a current switch are conductive and non-conductive, respectively. As a result, the base voltage of a transistor Q.sub.33 becomes low which tends to cause a decrease in the voltage V.sub.RW. Contrary to this, in the read mode, since the voltage of the write enable signal WE is higher than the reference voltage V.sub.R3, the transistors Q.sub.31 and Q.sub.32 are non-conductive and conductive, respectively. As a result, the base voltage of the transistor Q.sub.33 becomes high, which tends to cause an increase of the voltage V.sub.RW. Therefore, the base voltages of the transistors Q.sub.7 ' (FIG. 16) can be controlled by using the current switching circuit of FIG. 17.

FIG. 18 is a circuit diagram illustrating a fourth embodiment of the decoder/driver circuit of the present invention. The elements in FIG. 18 which are identical to those of FIGS. 7 and 9A are denoted by the same reference numerals as used in FIGS. 7 and 9A. As may be understood from FIG. 18, each of the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 is the same as that of FIG. 9A, except that a diode D.sub.13 which is controlled by the current switching circuit CS2 of FIG. 17 is provided. The output voltage V.sub.RW of the current switching circuit CS2 is applied to the anodes of the diodes D.sub.13, while the cathodes of the diodes D.sub.13 are connected to the bases of the transistors Q.sub.X. As explained before, the current switching circuit CS2 produces a signal whose voltage V.sub.RW is low in the write mode and high in the read mode. Therefore, the diode D.sub.13 is non-conductive in the write mode and conductive in the read mode. For example, in the non-selected word driver WD.sub.1 , the voltage V.sub.X in the write mode is determined by the voltage drop induced by the resistor R.sub.X. In addition, the voltage V.sub.X in the read mode is higher than that in the write mode, since the diode D.sub.13 is conductive. In other words, the voltage V.sub.X is clamped by the diode D.sub.13. On the other hand, for example, in the selected word driver WD.sub.0, the voltage V.sub.X is high regardless of the write or read mode. Thus, the decoder/driver circuit of FIG. 18 can be operated as illustrated in FIG. 5. It should be noted that emitter followers which serve as diodes can be used instead of the diodes D.sub.13. The embodiment of FIG. 18 can be also used in FIG. 10.

FIGS. 19 and 20 are circuit diagrams illustrating fifth and sixth embodiments of the decoder/driver circuit of the present invention. The decoder/driver circuits of FIGS. 19 and 20 are similar to that of FIG. 18. Each of the word drivers WD.sub.0, WD.sub.1, . . . , WD.sub.15 in FIG. 19 is based on those in FIG. 9B, while each of those in FIG. 20 is based on those in FIG. 9C. In addition, the operation of the circuits of FIGS. 19 and 20 is the same as that of FIG. 18.

In the above-mentioned embodiments, the number of the word lines (word drivers) is 16. However, it goes without saying that the present invention can be also applied to a bipolar memory where the number of word lines is other than 16 such as 32, 64, or 128.

As explained above, the system for driving a bipolar RAM according to the present invention has the following advantages, as compared with those of the prior art.

(1) High reading speed can be obtained, since the difference in potential between the selected word line and the non-selected word lines is set at the margin of the read operation.

(2) Stable reading operation can be obtained, since the data written into a memory cell cannot be destroyed during the read mode after the write mode.