Logic circuit with single charge pulling out transistor and semiconductor integrated circuit using the same

[0002] This is a division of application Ser. No. 09/327,834, filed Jun. 8, 1999 which is a continuation of application Ser. No. 08/783,026, filed Jan. 14, 1997 now abandoned.

[0003] (1) Field of the Invention

[0004] The present invention relates to logic circuits and semiconductor integrated circuits using the same, and more particularly to logic circuits which can be advantageously utilized in multiple input circuits, such as semiconductor memory decoder circuits, provided in an array form, and semiconductor integrated circuits using theame.

[0005] (2) Decription of the Related Art

[0006] As an example of the prior art multiple input logic circuit, a two-input NAND circuit comprising a bipolar transistor and a MOS transistor (this structure being hereinafter referred to as BiCMOS structure) will now be described. FIG. 1 is circuit diagram showing the NAND circuit. As shown, the NAND circuit comprises a logic circuit stage 1 having a CMOS transistor structure and an output stage 2 having bipolar/MOS transistor 710 structure. Specifically, the logic circuit stage 1 includes a parallel circuit of two p-channel MOS transistors M₂₃and M₂₄and a series circuit of two n-channel MOS transistors M₂₁and M₂₂, the parallel and series circuits being connected in series in the mentioned order between a high potential power supply line 3 (at a potential V_cc) and a ground line 4. Of the two input signals A and B subjected to logical operation, the signal A is inputted to the gates of the p- and n-channel MOS transistors M₂₃and M₂₁. The input signal B, on the other hand, is inputted to the gates of the p- and n-channel MOS transistors M₂₄and M₂₂. A signal which is produced as a result of the operation, is outputted from the common node of the drains of the two p-channel MOS transistors and the n-channel MOS transistor M₂₁and inputted to the output stage 2.

[0007] The output stage 2 includes bipolar transistor (hereinafter referred to as BiP transistor) Q₂and two n-channel MOS transistors M₂₅and M₂₆, these transistors being connected in series in the mentioned order between the power supply line 3 and the ground line 4. The logic signal from the logic circuit stage 1 is inputted to the base of the transistor Q₂. The input signals A and B are also assigned to the gates of the series n-channel MOS transistors M₂₅and M₂₆, respectively. This two-input NAND circuit having the BiCMOS structure provides its output signal X₀from an output terminal 5, which is constituted by the nodal point between the drain of the n-channel MOS transistor M₂₅and the emitter of the BiP transistor Q₂in the output stage 2.

[0008] In the circuit shown in FIG. 1, when the signals A and B both become “high”, the series n-channel MOS transistors M₂₁and M₂₂both become “on”, while the parallel p-channel MOS transistors M₂₃and M₂₄both become “off”. The base of the BiP transistor Q₂is thus brought down to the ground potential so that this transistor is turned off. At this time, the series n-channel MOS transistors M₂₅and M₂₆also both become “on” to cause the charge to a load (not shown) to be discharged through them, thus pulling down the output signal X₀to be “low”.

[0009] When either of the signals A and B becomes “low”, either of the n-channel MOS transistors M₂₁and M₂₂becomes “off”, and these transistors thus cannot pull down the base potential on the BiP transistor Q₂. Either of the p-channel MOS transistors M₂₃and M₂₄, on the other hand, is turned on, and these transistors thus pull up the base potential on the BiP transistor Q₂. The BiP transistor Q₂is thus turned on with its base potential pulled up to V_cc. Since either of the series n-channel MOS transistors M₂₅and M₂₆is turned off at this time, no charge is pulled through the output terminal 5. As the result of charging by the BiP transistor Q₂, the output signal X₀becomes “high”.

[0010] It is to be understood that the output X₀of the circuit shown in FIG. 1 has the NAND logic such that it becomes “low” when a plurality of inputs all become “high” and becomes “high” otherwise. The above logic circuit is used frequently for decoder circuits assembled in semiconductor memories or the like. The circuit operation in such a case is featured in that only one of a number of NAND gates in an array provides a “low” output as a selected output while the other NAND gate outputs are all “high” as non-selected outputs. The decoder circuit can finally select a memory cell corresponding to an inputted address with a connection of a plurality of stages of such NAND gates.

[0011] In the above NAND circuit, the output signal X₀is pulled up to “high” by the BiP transistor Q₂whose base current is supplied from at least either one of the p-channel MOS transistors M₂₃and M₂₄. Thus, high current capacity and high operation speed can be obtained. However, the output signal X₀is pulled down by the pull-down of the base potential on the BiP transistor Q₂by the series n-channel MOS transistors M₂₁and M₂₂and also by the pullout of charge from the output load by the series n-channel MOS transistors M₂₅and M₂₆. This is equivalent to doubling of the n-channel MOS transistor gate length and reduction to one half of the current capacity of the n-channel MOS transistor drain current. To compensate for the reduction to one half of the n-channel MOS transistor drain current, usually the gate width of the n-channel MOS transistors M₂₁, M₂₂, M₂₅and M₂₆is made large so as to increase the current capacity to prevent the speed-down of the pull-up as much as possible. However, by increasing the gate width, the input capacitance viewed from the input signal is increased resulting in a delay in the operation of a preceding logic circuit stage (not shown). Where a plurality of logic gate stages are present, it is necessary for the high speed operation of the circuit to improve the fan-out characteristic, i.e., the relation between the ratio of the input capacitance to the output load capacitance and the delay time, but this has been seriously impeded by the capacity reduction of the series n-channel MOS transistors M₂₁, M₂₂, M₂₅and M₂₅.

[0012] In the decoder circuit, the operation speed of the memory cell selection which is the purpose of the circuit, is greatly dependent on the delay time in the selected signal output. The selection is effected for the pull-down of the output by the series n-channel MOS transistors M₂₁and M₂₂and also M₂₅and M₂₆(or for the pull-up of the output by series p-channel MOS transistors in a NOR circuit). Therefore, the operation speed is greatly influenced by the reduction of the current capacity due to the series connection of MOS transistors.

[0013] An object of the invention, therefore, is to overcome the problems existing in the prior art, and to provide a logic circuit in which the number of series connected MOS transistors constituting the route for supplying charge to the load and pulling out charge therefrom is reduced and which is capable of operating at a high speed and is small in lay-out area.

[0014] Another object of the invention is to provide a logic circuit which is applicable to such circuits as semiconductor memory decoder circuits using a number of logic circuits disposed in an array and which enables the high selection speed.

[0015] According to one aspect of the invention, there is provided a logic circuit performing a predetermined logic operation by supplying charge to an external load or pulling out charge therefrom according to a combination of the states of a plurality of externally inputted binary signals, the logic circuit comprising:

[0016] at least a first transistor for supplying charge through an output terminal to the external load; and

[0017] at least a second transistor for pulling out the charge from the load through the output terminal,

[0018] one of the first and second transistors being constituted by a MOS field-effect transistor having a drain connected to the output terminal,

[0019] the MOS field-effect transistor having a source receiving an inverse signal inverse to a signal combined for logic operation with an input signal inputted to a gate of the MOS field-effect transistor.

[0020] The logic circuit may further comprises an inverse signal generating circuit for generating an inverse signal inputted to the source of the MOS field-effect transistor,

[0021] the inverse signal generating circuit for generating the inverse signal, in which the logic amplitude thereof is reduced according to a down-threshold in two series n-channel MOS field-effect transistors connected between a power supply line and a reference potential point, by inputting complimentary signals in phase and in inverse phase with respect to the inverse signal to the gates of the n-channel MOS field-effect transistors, respectively.

[0022] According to another aspect of the invention, there is provided a semiconductor integrated circuit comprising a decoder circuit provided on a chip, the decoder circuit having an array of a plurality of logic circuits performing a predetermined logic operation by supplying charge to an external load or pulling out charge therefrom according to a combination of the states of a plurality of externally inputted binary signals, each of the logic circuits comprising:

[0023] at least a first transistor for supplying charge through an output terminal to the external load; and

[0024] at least a second transistor for pulling out the charge from the load through the output terminal, one of the first and second transistors being constituted by a MOS field-effect transistor having a drain connected to the output terminal,

[0025] the MOS field-effect transistor having a source receiving an inverse signal inverse to a signal combined for logic operation with an input signal to a gate of the MOS field-effect transistor,

[0026] the MOS field-effect transistor being arranged such that the adjacent ones of the logic circuits share a source diffusion layer.

[0027] The above and other objects, features and advantages of the present invention will be apparent from the following description of preferred embodiments of the invention explained with reference to the accompanying drawings, in which:

[0028] FIG. 1 is a circuit diagram showing a prior art NAND circuit;

[0029] FIGS. 2A, 2B and 2C are respectively a circuit diagram, a truth table and a layout pattern of a NAND circuit of a first embodiment according to the invention;

[0030] FIG. 3 is a circuit diagram showing a NAND circuit of a second embodiment according to the invention;

[0031] FIG. 4 is a circuit diagram showing a NAND circuit of a third embodiment according to the invention;

[0032] FIG. 5 is a graph showing the relation between the delay time and the number of elements in a prior art NAND circuit and a NAND circuit having a BiCMOS structure embodying the invention;

[0033] FIG. 6 is a circuit diagram showing a NOR circuit of a fourth embodiment according to the invention;

[0034] FIG. 7 is a circuit diagram showing a NOR circuit of a fifth embodiment according to the invention;

[0035] FIG. 8 is a circuit diagram showing an AND circuit of a sixth embodiment according to the invention;

[0036] FIG. 9 is a circuit diagram showing an AND circuit of a seventh embodiment according to the invention;

[0037] FIG. 10 is a circuit diagram showing an AND circuit with an inverse signal generator of an eighth embodiment according to the invention;

[0038] FIG. 11 is a circuit diagram showing an AND circuit of a ninth embodiment according to the invention;

[0039] FIG. 12 is a circuit diagram showing an OR circuit of a tenth embodiment according to the invention; and

[0040] FIG. 13 is a graph showing the relation between the delay time and the C_out/C_inof a prior art AND circuit and AND circuits having CMOS structure embodying the invention.

[0041] Now, preferred embodiments of the invention are explained with reference to the drawings.

[0042] FIG. 2A is a circuit diagram showing a first embodiment of the logic circuit according to the invention applied to a two-input NAND logic circuit having the BiCMOS structure. Comparing FIGS. 2A and 1, this embodiment of the invention is significantly different from the prior art NAND circuit in that the current paths for pulling down output stage BiP transistor base potential and pulling out charge from the load are constituted by independent n-channel MOS transistors M₁₁and M₁₄, respectively, instead of the series n-channel MOS transistors M₂₁and M₂₂and the series n-channel MOS transistors M₂₅and M₂₆in the prior art.

[0043] In this embodiment, an input signal A is inputted to the gate of a p-channel MOS transistor M₁₂as well as the gate of the n-channel MOS transistor M₁₁. The source of the p-channel MOS transistor M₁₂is connected to a high potential power supply line 3 (at V_cc). A separate input signal B is inputted to the gate of a p-channel MOS transistor M₁₃which is parallel with the transistor M₁₁. The two p-channel MOS transistors have a common drain, which is connected to the base of a BiP transistor Q₁with the collector thereof connected to the power supply line 3. The emitter of the BiP transistor Q₁and the drain of the n-channel MOS transistor M₁₄, to which the signal A is a gate input, are connected together, and this node is connected to an output terminal 5. The inverse signal ∇B inverse to the input signal B is inputted to the sources of the n-channel MOS transistors M₁₁and M₁₄(VB here and hereinafter is substituted for an upper bar indicating a signal inverted).

[0044] The logic operation of this embodiment will now be described with reference to the truth table shown in FIG. 2B. When the input signal A becomes “low” (i.e., “0”), the n-channel MOS transistors M₁₁and M₁₄are turned off, while the p-channel MOS transistor M₁₂is turned on. The BiP transistor Q₁is thus turned on with its base potential pulled up to the supply voltage V_ccirrespective of the signal B. A large current is thus caused to flow from the collector to the emitter to charge the load (not shown) connected to the output terminal 5. The output X₀thus becomes “high” (i.,e., “1”).

[0045] When the input signal A becomes “high”, the state of output X₀changes depending on the state of the input signal B. When the signal B is “low” at this time, the inverse signal ∇B is “high”. The n-channel MOS transistors M₁₁and M₁₄are thus turned off. The p-channel MOS transistor M₁₂is also turned off, but the other p-channel MOS transistor M₁₃is turned on. As in the above case, the output X₀thus becomes “high”. When the input signal B is “high”, the n-channel MOS transistors M₁₁and M₁₄both become “on” since their gate and source potentials become “high” and “low”, respectively. The p-channel MOS transistors M₁₂and M₁₃, on the other hand, both become “off”. The BiP transistor Q₁is thus turned off with its base potential brought down to the ground potential. The load is thus discharged by the “on” current capacity of the n-channel MOS transistor M₁₄, thus pulling down the output X₀to the “low” level.

[0046] The transient response of this embodiment is featured by the operation of input to the sources of the n-channel MOS transistors M₁₁and M₁₄with the inverse signal ∇B. The source capacitance of the n-channel MOS transistor M₁₄used in this embodiment will now be described. FIG. 2C is a layout pattern concerning the n-channel MOS transistor M₁₄in the circuit shown in FIG. 2A. This layout pattern assumes a decoder circuit of a semiconductor memory device. Where there are four NAND gates NAND(1) to NAND(4) in an array, the layout of the n-channel MOS transistor M₁₄in each of the gates is as shown. The inverse signal ∇B is shared by the four NAND gates. As the input signal A, signals A₁to A₄are each inputted to the gate of the n-channel MOS transistor M₁₄in each of the NAND gates. In this case, it is important that a source diffusion layer is shared by the two NAND n-channel MOS transistors M₁₄in two NAND gates. With this arrangement, the capacitance of the source diffusion layer of the n-channel MOS transistor M₁₄is reduced to one half. Particularly, with a device with a large side surface capacitance on the field side, it is reduced to be less than one half. With this structure of this embodiment, the input capacitance is reduced to about one half the usual gate input capacitance, thus improving the fan-out characteristics. In addition, the n-channel MOS transistor M₁₄functions as a transfer gate and starts pulling down the current from the output terminal 5 sooner than it is turned on by its gate input. It is thus possible to obtain a very fast operation inclusive of the operation of a preceding stage logic circuit.

[0047] It has to be noted, however, that the load capacitance at the output terminal 5 is discharged through the line of the inverse signal ∇B as an input signal, and the following condition (1) has to be met for the fast operation.

[0048] Load capacitance at output terminal 5<Load capacitance with respect to inverse signal ∇B  (1)

[0049] When the load capacitance with respect to the inverse signal ∇B is high, the influence of the discharge current from the output terminal 5 is negligible. While the above example is the case when source diffusion layer is shared by adjacent NAND gates of n-channel MOS transistors M₁₄, the advantages of such a layout structure are also obtainable with n-channel MOS transistors M₁₁.

[0050] The response to the input of the signal A will now be considered. The sources of both the n-channel MOS transistors M₁₁and M₁₄are connected to the ground line. That is, the output X₀is pulled down by the single stage n-channel MOS transistor. Thus, fast operation is possible again. The input of the inverse signal ∇B to the n-channel MOS transistor source does not cause the source voltage to float significantly so long as the above condition (1) of the load capacitance ratio of the output terminal 5 is being satisfied, and it is possible to pull out the high n-channel MOS transistor current capacity.

[0051] In the case of an actual decoder circuit, the signal B and the inverse signal ∇B inverse thereto are or must be readily supplied simultaneously. In the case of four NAND circuits with two address signals inputted thereto, signals in phase with the two address signals and the inverse signals inverse to the in-phase signals are input signals, and are necessarily present when a decoder circuit is assumed. In decoder circuits following a first decoder circuit, the inverse signals may not necessarily be present. Even in such a case, it is readily possible to generate inverse signals for a plurality of NAND gates via an inverter stage. In this case, it is desirable to generate the signal B which has a small load.

[0052] A second embodiment of the invention will now be described with reference to FIG. 3. In this embodiment, only a single p-channel MOS transistor M₃₂is used for the pull-up of the base potential on an output stage BiP transistor Q₃. The p-channel MOS transistor M₃₂is normally held “on” with its gate held at the ground potential from a ground line 4. The input signals are thus only the signal A and the inverse signal and, unlike in the preceding first embodiment, the inverse signal ∇B is unnecessary. This embodiment also has a lesser number of elements than that in the first embodiment by one.

[0053] In this embodiment, in the other cases than when the signal A and the inverse signal ∇B come up with “high” and “low”, respectively, p-channel MOS transistors M₃₁and M₃₂are “off”. Thus, the BiP transistor Q₃is held “on” with its base potential held pulled up to the supply voltage V_ccby the p-channel MOS transistor M₃₂. As a consequence, the Bip transistor Q₃turns to an “on” state so that the output X₀is made “high”. When the signal A and the inverse signal ∇B are “high” and “low”, respectively, the n-channel MOS transistors M₃₁and M₃₃both become “on”, and the output X₀is pulled down to the “low” level by the current capacity of the n-channel MOS transistor M₃₃.

[0054] Since the p-channel MOS transistor M₃₂is normally held “on”, the base potential on the BiP transistor Q₃is determined by the impedance ratio between the n- and p-channel MOS transistors M₃₁and M₃₂. The BiP transistor Q₃thus can be turned off by setting such that its base potential becomes lower than about 0.8 V which is the forward voltage across its base-emitter p-n junction. In this case, no problem arises in the “low” output. However, the size of the p-channel MOS transistor M₃₂cannot be made so large. This is so because the size determines the base potential on the BiP transistor Q₃in the “off” state thereof in addition to the “on” capacity thereof. Therefore, the “high” output, i.e., the non-selected output of a decoder, is not provided so fast as in the first embodiment. The fan-out characteristic, however, is not substantially different since the input capacitance is reduced by that corresponding to a p-channel MOS transistor.

[0055] A third embodiment of the invention will now be described with reference to FIG. 4. Referring to the figure, this embodiment is the same as the preceding embodiments in the point wherein the output X₀is pulled down by using a single n-channel MOS transistor M₄₁. A feature of this embodiments resides in that the output X₀is pulled up by using only BiP transistors. The inverse signal ∇B which is inputted to the source of the n-channel MOS transistor M₄₁, is also inputted to the base of the BiP transistor Q₄₁, and the inverse signal ∇A inverse to the signal A is inputted to a base of a BiP transistor Q₄₂which is parallel with the transistor Q₄₁.

[0056] In this embodiment, when the signal A and the inverse signal ∇B become “high” and “low”, respectively (i.e., A=1, B=1), the n-channel MOS transistor M₄₁is turned on. At this time, the inverse signals ∇A and ∇B both become “low”, and the two BiP transistors Q₄₁and Q₄₂are both turned off to provide the output X₀as a “low” output. In the other cases, the n-channel MOS transistor M₄₁is “off”, while at least one of the BiP transistors Q₄₁and Q₄₂is turned on, so that the output X₀is pulled up to the “high” level. In this embodiment, fast operation can be expected, since the output pull-up BiP transistors Q₄₁and Q₄₂are driven directly by the inverse signals ∇A and ∇B. However, since the BiP transistor base capacitances can be seen from the input signals, the fan-out characteristic is not improved so much as in the preceding embodiments. However, the use of only three elements in this embodiment is a great merit in view of, for instance, the layout area on the chip.

[0057] FIG. 5 shows the performances of the first and third embodiments of the invention in comparison with the performance of the prior art NAND circuit. In FIG. 5, the ordinate is taken for the delay time in the NAND circuit inclusive of the inverter connected thereafter, and the abscissa is taken for the number of elements in the NAND circuit. The delay time is taken by setting the ratio C_out/C_inof the output load capacitance C_outto the input capacitance C_into a constant, i.e., 20, from the consideration of the fan-out characteristic. As the delay time, the average values of the two inputs when they are respectively “high” and “low” are compared. The first embodiment has elements lesser by two than those in the prior art, while the delay time in this case is improved by nearly 20% from that in the prior art. With the third embodiment, the delay time is improved only by about 10% from that in the prior art, but the number of elements is reduced to be less than one half. Selection as to whether to attach importance to the operation speed improvement or to the layout area reduction by reducing the number of elements, may depend on the desired performance of the product to be designed.

[0058] A fourth embodiment of the invention will now be described with reference to FIG. 6. This embodiment is a two-input NOR circuit having the BiCMOS structure. Referring to FIG. 6, the base potential on a BiP transistor Q₆is pulled down by parallel n-channel MOS transistors M₆₂and M₆₃, to the gates of which input signals A and B are inputted. Transistors M₆₄and M₆₅are for pulling down an output X₁. For the pull-up of the base potential on the BiP transistor Q₆, a single p-channel MOS transistor M₆₁is provided, to the gate and source of which the signal A and the inverse signal ∇B are inputted, respectively. In the prior art NOR circuit, this part is constituted by two series p-channel MOS transistors. In such a case, there is a problem in that, when providing the “high” output, the current capacity of the p-channel MOS transistors in the “on” state thereof is low and unable to supply sufficient base current to the transistor Q₆. In other words, the capacity of the BiP transistor Q₆cannot be made full use of, and it is not readily enables the high speed operation. With this embodiment of the NOR circuit, like the case of the NAND circuit, the current capacity of the p-channel MOS transistor can be increased thus enabling the high speed operation.

[0059] A fifth embodiment of the invention will now be described with reference to FIG. 7. Unlike the preceding fourth embodiment (see FIG. 6), in this embodiment only a single n-channel MOS transistor M₇₂constitutes a base potential pull-down transistor on an output stage BiP transistor Q₇. This n-channel MOS transistor M₇₂is normally held “on” with its gate held at the supply voltage V_ccsupplied from the power supply line 3. In this circuit construction, the base potential on the BiP transistor Q₇can be readily pulled up to the supply voltage V_ccso long as the n-channel MOS transistor M₇₂can be turned off. In this embodiment, since the transistor M₇₂is normally held “on”, the base potential on the transistor Q₇is lower than the supply voltage V_ccby an amount corresponding to the impedance ratio between the p- and n-channel MOS transistors M₇₁and M₇₂. This base potential reduction appears as a reduction of the output X₁, and it should be made to an extent that it does not deteriorate the circuit characteristics of the next stage.

[0060] In the NOR circuit of the preceding fourth and fifth embodiment, the output of the p-channel MOS transistor M₇₁which receives its input at its source, is coupled to the base of the BiP transistor Q₇. Thus, as compared to the NAND circuit, the condition (1) concerning the ratio between the output load capacitance and the load capacitance of the inverse signal ∇B which is a condition for the high speed operation, can be readily attained. This is so because the output X₁is changed via the BiP transistor Q₇. The operation speed, however, is not improved so much as in the NAND circuit. In addition, since it is only a single p-channel MOS transistor that can receive the input thereto as the source, the element number reduction effect is not so much as in the NAND circuit.

[0061] A sixth embodiment of the invention will now be described with reference to FIG. 8. This embodiment of the logic circuit is a CMOS structure AND circuit, which comprises a NAND circuit and an inverter and does not use any BiP transistor. Referring to FIG. 8, input signals A and B are assigned to the gates of parallel p-channel MOS transistors M₈₂and M₈₃, respectively. Also, the signal A and the inverse signal ∇B are inputted to the gate and source, respectively, of the n-channel MOS transistor M₈₁. The drains of the p- and n-channel MOS transistors are connected so as to constitute a common node, and a NAND output therefrom is inputted to a CMOS inverter which includes a p- and an n-channel MOS transistor M₈₄and M₈₅in series with each other. The inverter output is the output X₂of the AND circuit.

[0062] In this embodiment, the NAND logic part is the same in the circuit operation and characteristics as the previous NAND circuit having the BiCMOS structure (i.e., the first embodiment shown in FIGS. 2A to 2C). In this embodiment, however, the AND logic form comprising the NAND circuit and the CMOS inverter connected to the output side of the NAND circuit constitutes a basic unit. In the CMOS circuit, the drive capacity with respect to the load is low compared to the BiCMOS circuit. Accordingly, for driving a large load, it is general to connect a drive inverter to a plurality of circuits, i.e., logic circuits. This way of driving is suitably adoptable for operations with an n-channel MOS transistor receiving the inverse signal ∇B at the source as well. As already described, for fast operation, it is necessary that the NAND output load capacitance be lower than the load capacitance with respect to the inverse signal ∇B. This requirement, however, may fail to be satisfied in the case of driving a large load of the output X₂directly with the NAND output. In a decoder circuit, its output is inputted to a number of next stage logic circuits. Most of these circuits are not selected and, only in selected circuits, the NAND output load is seen through the source of the n-channel MOS transistor M₈₁. The output load capacitance of the NAND output part in this case is constituted by the sole driving CMOS inverter input capacitance. The load capacitance of most of the decoder outputs is thus the sum of the non-selected circuit side input capacitance and the inter-connection capacitance. It is thus possible to readily meet the requirement for the fast operation, and high speed performance may be readily obtained.

[0063] A seventh embodiment of the invention will now be described with reference to FIG. 9. In this embodiment, a single p-channel MOS transistor M₉₂which is normally held “on”, is used in lieu of the p-channel MOS transistors in the NAND circuit part of the CMOS structure AND circuit as in the above sixth embodiment, while a CMCS inverter is connected to the NAND circuit to provide an AND output. With this construction, when the n-channel MOS transistor M₉₁is turned on, the NAND output is not lowered down to the ground potential but is slightly floated thereabove.

[0064] When the extent of the floating is less than the threshold voltage of a subsequent inverter stage n-channel MOS transistor M₉₄, however, no great through current will be caused. Of course, a through current through the n-channel MOS transistor M₉₂is caused. However, this transistor is small in size and low in current capacity. In the case of a decoder circuit, the through current flows through only selected circuits, which are extremely small number, so that this does not pose any substantial problem.

[0065] An eighth embodiment of the invention will now be described with reference to FIG. 10. In the preceding seventh embodiment AND circuit (see FIG. 9), the inverse signal ∇B is supplied externally to the logic circuit. This embodiment features that it includes a circuit for generating the inverse signal ∇B. In this embodiment, a plurality of AND circuits (outputs X₂₁to X_2n) are driven by an inverse signal generator, which is a series connection of a pull-up n-channel MOS transistor M₁₀₃and a pull-down n-channel MOS transistor M₁₀₄. This inverse signal generator functions as a push-pull type circuit to drive as signal line 6 of the inverse signal ∇B with signals ∇Band B₀in phase and in inverse phase respectively with the inverse signal ∇B being inputted to the gates of the n-channel MOS transistors M₁₀₃and M₁₀₄, respectively.

[0066] Of the MOS transistors in the AND circuit shown in FIG. 10, the n-channel MOS transistor M₁₀₂, to the source of which the inverse signal ∇B is inputted, is turned off when its source potential becomes lower than its gate potential by its threshold voltage V_TH. This means that in this embodiment the source potential on the n-channel MOS transistor M₁₀₂, i.e., “high” level of the inverse signal ∇B, need be pulled up only to V_cc-V_TH. The inverse signal ∇B is pulled up to the “high” level by the n-channel MOS transistor M₁₀₃, that is, it is pulled up to the “high” level when the input signal ∇B₀to the inverted signal generator becomes “high”. At this time, the “high” level of the output signal ∇B₀that is obtained is V_cc-V_THdue to the “threshold pull-down” in the n-channel MOS transistor M₁₀₃although the “high” level of the input signal ∇B₀is equal to the supply voltage V_cc. This “high” level of the inverse signal ∇B coincides with the turn-off level of the AND logic part n-channel MOS transistor M₁₀₂. With the construction of this arrangement, it is possible to reduce the amplitude of the inverse signal ∇B and obtain both the effects of speeding up the operation and reducing the current consumption. For example, assuming the supply voltage V_ccto be 3.3 V, the delay time up to the appearance of the AND output can be reduced by about 20%, while the current consumption can be reduced by about 30%. It is further possible, by an n-channel MOS transistor in lieu of the low current capacity p-channel MOS transistor for the inverse signal generator, to reduce the transistor size and reduce the layout area of the inverse signal drive line buffer by about 25%.

[0067] A ninth embodiment of the invention will now be described with reference to FIG. 11. This embodiment is featured in that, as compared with the previous seventh embodiment, a p-channel MOS transistor M₁₁3 is connected in parallel with a p-channel MOS transistor M₁₁₂which is normally held “on”. The inverse signal inverse to the AND output X₂is inputted to the gate of the p-channel MOS transistor M₁₁₃. When the NAND output is caused to go to “high” from “low” by this feed-back, the p-channel MOS transistor M₁₁₃is turned on and cooperates with the transistor M₁₁₂to supply charge, thus quickly raising the NAND output. When the NAND output is pulled down from “high” to “low”, the p-channel MOS transistor M₁₁₃is turned off and blocks current to an n-channel MOS transistor M₁₁₁. Thus, it has no influence on the NAND output pull-down speed. The effect of the feed-back for the high speed operation is enhanced by setting the current capacity of the p-channel MOS transistor M₁₁₃to be high with respect to the current capacity of the p-channel MOS transistor M₁₁2. In this case, however, care should be taken for the floating of the “low” level of the NAND output and the reduction of the capacity of retaining the “high” level of this output.

[0068] A tenth embodiment of the invention will now be described with reference to FIG. 12. This embodiment is an OR gate, which comprises a NOR gate part having a series connection between a p-channel MOS transistor M₁₂₁and an n-channel MOS transistor M₁₂₂, and an inverter 7 for inverting the NOR output. As in the n-channel MOS transistor in the above AND circuit (i.e., the seventh embodiment shown in FIG. 9), the signal A is inputted to the gate of the p-channel MOS transistor M₁₂₁, while inputting the inverse signal ∇B to the source of this transistor. The p-channel MOS transistor M₁₂₁is turned on when and only when the signal A and the inverse signal ∇B are “low” and “high” (i.e., A=0, B=0), respectively. An output X₃thus becomes “low” only at this time. Otherwise, the output X₃is “high” since the p-channel MOS transistor M₁₂₁is normally “off”. With this embodiment, as in the preceding embodiments, fast operation can be readily realized.

[0069] FIG. 13 shows the MOS transistor AND circuit fan-out characteristics of the prior art AND circuit and the embodiments of the invention. In the figure, the ordinate is taken for the delay time, and the abscissa is taken for C_out/C_inin the AND circuit comprising the NAND circuit and the inverter. The delay time as shown is the average delay time when the two inputs are inverted between “high” and “low”. Regarding the prior art AND circuit, values in the general CMOS structure circuit are shown while, regarding the invention, values in the AND circuits as the seventh and ninth embodiments are shown. It will be seen from the figure that the operation is faster with the embodiments of the invention, particularly with the ninth embodiment incorporating the feed-back, than with the prior art circuit irrespective of the fan-out coefficients. Specifically, with the same C_out/C_inthe delay time can be reduced by about 30 to 40%.

[0070] As has been described in the foregoing, according to the invention, the charge pull-out route in the NAND logic circuit which has heretofore been constituted by a series connection of a plurality of N-channel MOS transistors, is constituted by a single n-channel MOS transistor in which, in addition to a signal inputted to its gate, the inverse signal is directly inputted also to its source. The single n-channel MOS transistor permits increasing the output pull-down capacity. In the case of a BiCMOS NAND circuit, this construction can be adopted for both the output pull-down and the output stage bipolar transistor base potential pull-down to increase the operation speed, reduce the number of elements and reduce the layout area. In the case of applying the above structure to a CMOS NAND circuit, an AND circuit together with an inverter is made to be a basic unit whereby the NAND output load capacitance with respect to the source input inverse signal line capacitance is reduced and the operation speed is increased.

[0071] Further increase of the operation speed and reduction of the power consumption can be obtained by compressing the amplitude of the source input inverse signal by the provision of a circuit for generating the inverse signal.

[0072] The invention is applicable to such logic circuits as NOR circuits and OR circuits as well as NAND circuits and AND circuits, and when applied particularly to semiconductor memory decoder circuits or the like, it permits obtaining great effects of increasing the operation speed, reducing the layout area and reducing the power consumption.

[0073] While the invention has been described in its preferred embodiments, it is to be understood that the words which have been used are words of description rather than limitation and that changes within the purview of the appended claims may be made without departing from the true scope of the invention as defined by the claims.