Logic circuit for fast carry/borrow

1. Field of the Invention

The present invention relates generally to a logic circuit for fast carry or borrow, more particularly, to a logic circuit for use in an incrementer or decrementer having a transfer gate chain for carry or borrow propagation.

2. Description of the Related Art

FIG. 11 shows a prior art ripple carry type incrementer.

This incrementer asynchronously adds a bit C 1 =“1” to 4 bit inputs A 1 to A 4 to obtain 5 bit outputs D 1 to D 4 and C 5 . A circuit for each digit is a half adder, and each half adder has the same configuration. A half adder circuit 10 for the least significant digit consists of: an exclusive-OR gate 11 whose output bit D 1 takes on ‘1’ when either input bit A 1 or C 1 is ‘1’ and the other is ‘0’; and an AND gate 12 whose output bit C 2 as a carry-out bit to the upper digit takes on ‘1’ when the both input bits A 1 and C 1 are ‘1’.

Although such a ripple carry type circuit is simple in configuration and can be down-sized on circuit scale, since a carry-in from the lower digit at each digit is logically operated by an AND gate, determination of a carry-out bit C 5 is delayed, resulting in a low speed operation. For example, when the input bits A 4 to A 1 =‘1111’ is provided in a state C 1 =‘1’, carries C 2 to C 5 sequentially change to ‘1’, and therefore determination of the output value ‘10000’ is delayed.

FIG. 12 shows a binary carry logic circuit 12 A for use in a full adder disclosed in JP 05-61645 A.

This circuit 12 A consists of: transfer gates 13 to 15 ; and inverters 16 to 18 . When A 1 =‘0’, the transfer gates 13 and 14 are on and off, respectively to be D 1 =C 1 . When A 1 =‘1’, the transfer gates 13 and 14 are turned off and on, respectively to be D 1 =*C 1 , where the symbol * denotes an inverse operator. From these relations, the transfer gates 13 and 14 and the inverters 16 and 17 constitutes an exclusive-OR gate 11 A.

When the output bit D 1 ‘0’, that is, when the input bit A 1 and the carry-in bit C 1 from the lower digit are both ‘1’, the transfer gate 15 is on to be C 2 =A 1 .

According to such a binary carry logic circuit 12 A, since the input bit A 1 passes through the transfer gate 15 to be the carry-out bit C 2 to the upper digit, it seems that the operation is fast. However, the carry-in bit C 1 , for example, turns on the NMOS transistor of the transfer gate 15 through the inverter 16 , the transfer gate 14 and then the inverter 18 , which is an obstacle against a high speed operation.

FIG. 13 shows a prior art ripple carry type decrementer.

A half subtractor circuit 10 X for the least significant digit is of the same configuration as the half adder circuit 10 with the exception that an inverter 19 is connected between the input bit A 1 and one input of the AND gate 12 . A borrow B 2 , which is an output of the AND gate 12 , takes on ‘1’ when the input bit A 1 =‘0’ and a borrow B 1 =‘1’.

Although this decrementer is also of a simple configuration like the incrementer of FIG. 11 to enable a circuit scale to be downsized, since a borrow-in bit from the lower digit at each digit is logically operated in an AND gate, determination of a borrow B 5 is delayed, resulting in a low speed operation.

On the other hand, a carry look ahead type incrementer and a carry look ahead type decrementer are faster in operation than those of a ripple carry type. However, the circuit scale thereof is larger.

In such a way, in regard to an incrementer and a decrementer, there is a trade-off relation between a high speed operation and downsizing on circuit scale.

Accordingly, it is an object of the present invention to provide a logic circuit for fast carry or borrow capable of achieving a high speed operation while maintaining an advantage of a ripple carry type with a small circuit scale.

In one aspect of the present invention, there is provided an incrementer comprising a plurality of half adder circuits each adding a carry-in bit to an input bit to generate an output bit and a carry-out bit, the plurality of half adder circuits being connected in cascade in regard to the carry-in and carry-out bits. Each of the half adder circuits other than one for the least significant digit comprises: a transfer gate, having a data input and a data output, turned on when the input bit is active, the data input receiving the carry-in bit; a transistor, having a current channel connected between a power supply potential and the data output, a logic value of the power supply potential being equal to that of the carry-in bit in an inactive state, turned on when the input bit being inactive; and a logic circuit, generating the output bit which is active when either the input bit or the carry-in bit is active; wherein the carry-out bit is on the data output.

With this configuration, since the transfer gates of the half adder circuits other than one for the least significant digit are connected in series to each other and the transfer gates are simultaneously on/off controlled by input bits, a carry bit from the least significant digit can propagate through a transfer gate chain at a high speed in the worst case.

In another aspect of the present invention, there is provided a decrementer comprising a plurality of half subtractor circuits each subtracting a borrow-in bit from an input bit to generate an output bit and a borrow-out bit, the plurality of half subtractor circuits being connected in cascade in regard to the borrow-in and borrow-out bits. Each of the half subtractor circuits other than one for the least significant digit comprising: a transfer gate, having a data input and a data output, turned on when the input bit is inactive, the data input receiving the borrow-in bit; a transistor, having a current channel connected between a power supply potential and the data output, a logic value of the power supply potential being equal to that of the borrow-in bit in an inactive state, turned on when the input bit being active; and a logic circuit, generating the output bit which is active when either the input bit or the borrow- in bit is active; wherein the borrow-out bit is on the data output.

With this configuration, since the transfer gates of the half subtractor circuits other than one for the least significant digit are connected in series to each other and the transfer gates are simultaneously on/off controlled by input bits, a borrow from the least significant digit can propagate through a transfer gate chain at a high speed in the worst case.

Other aspects, objects, and the advantages of the present invention will become apparent from the following detailed description taken in connection with the accompanying drawings.

FIG. 1 is a circuit diagram showing a half adder of a first embodiment according to the present invention;

FIG. 2 is a circuit diagram showing a 4 bit incrementer to which the half adder circuit of FIG. 1 is applied;

FIG. 3 is a layout representation of the circuit of FIG. 2;

FIG. 4 is a block diagram showing a 16 bit incrementer constructed of four 4-bit incrementers, each having the configuration of FIG. 2, connected in cascade;

FIG. 5 is a circuit diagram showing a half adder of a second embodiment according to the present invention;

FIG. 6 is a circuit diagram showing an example of FIG. 5 wherein the exclusive OR gate is constructed using transfer gates;

FIG. 7 is a circuit diagram showing a 4 bit incrementer to which the half adder circuit of FIG. 6 is applied;

FIG. 8 is a circuit diagram showing an incrementer of a third embodiment according to the present invention;

FIG. 9 is a circuit diagram showing a decrementer of a fourth embodiment according to the present invention;

FIG. 10 is a circuit diagram showing a decrementer of a fifth embodiment according to the present invention;

FIG. 11 is a logic circuit diagram showing a prior art ripple carry type incrementer;

FIG. 12 is a diagram showing a prior art binary carry logic circuit using transfer gates; and

FIG. 13 is a logic circuit diagram showing a prior art ripple carry type decrementer.

Referring now to the drawings, wherein like reference characters designate like or corresponding parts throughout several views, preferred embodiments of the present invention are described below.

First Embodiment FIG. 1 shows a half adder circuit of the first embodiment according to the present invention for use in, for example, an incrementer or an adder.

A binary carry logic circuit 20 consists of a transfer gate 21 , an inverter 22 , and a PMOS transistor 23 , and has a configuration simpler than the binary carry logic circuit 12 A of FIG. 12 .

The transfer gate 21 is a pair of an NMOS transistor and a PMOS transistor connected in parallel with each other, the NMOS transistor receives an input bit A 1 at its gate, and the PMOS transistor receives, at its gate, a signal *A 1 obtained by inverting the input bit A 1 with the inverter 22 , where the symbol* denotes a reverse operator and a signal indicated by a reference character having * is active-low, which applies in the following description as well. A carry-in bit *C 1 from the lower digit is provided to the data input of the transfer gate 21 . The PMOS transistor 23 is connected between the data output of the transfer gate 21 and a power supply potential VDD, and receives the input bit A 1 at its gate. A carry-out bit *C 2 to the upper digit is the signal on the data output of the transfer gate 21 .

In the above configuration, when A 1 =‘1’, the transfer gate 21 and the PMOS transistor 23 are on and off, respectively, to be *C 2 =*C 1 . When A 1 =‘0’, the transfer gate 21 and the PMOS transistor 23 are off and on, respectively, to be *C 2 =‘1’ regardless of the value of the carry-in bit *C 1 . Accordingly, *C 2 =0 only when A 1 =‘1’ and *C 1 =‘0’, and *C 2 =‘1’ in the other cases. That is, the following logic equation holds:

C 2 =A 1 ·C 1 .  (1)

Features of the binary carry logic circuit 20 in configuration are that on/off of the transfer gate 21 is reverse to that of the PMOS transistor 23 , and that the logic value of the source potential, VDD, of the PMOS transistor 23 is equal to that of the carry-in bit *C 1 which is in an inactive state.

In a logic circuit 30 , the carry-in bit *C 1 and the input signal *A 1 are provided to an AND circuit 31 , the carry-out bit *C 2 to the upper digit is provided to an inverter 32 , the outputs of the AND gate 31 and the inverter 32 are provided to a NOR gate 33 , and the output bit D 1 is taken out from the NOR gate 33 . The operation of the logic circuit 30 is represented by the following logic equation: D1 =      *  ( *  C1 ·   *  A1 + C2 ) =   *  ( *  C1 ·   *  A1 ) ·   *  C2 = ( C1 + A1 ) ·   *  C2 . 

The above equation is transformed to the following equation using the equation (1): D1 = ( C1 + A1 ) · (   *  A1 +   *  C1 ) = C1 ·   *  A1 +   *  C1 · A1 .  ( 2 )

Therefore, the output bit D 1 is ‘1’ when one of the input bit A 1 or the carry-in bit C 1 is ‘1’ and the other is ‘0’, while being ‘0’ in the other cases.

With such operation, the circuit of FIG. 1 functions as a half adder.

FIG. 2 is a circuit diagram showing a 4 bit incrementer to which the half adder circuit of FIG. 1 is applied. Each of half adder circuits for the upper three digits is of the same configuration as that of FIG. 1 . Transfer gates 212 to 214 of binary carry logic circuits 20 for the respective upper three digits are connected in series to each other. It is when input bits A 1 to A 4 are all ‘1’ that carry propagation is delayed most. In this case, since the transfer gates 212 through 214 are simultaneously turned on, a carry bit *C 2 from the least significant digit can propagate through the transfer gates 212 to 214 at a high speed. In order to accelerate this propagation to a higher speed, it is enough to steepen the slope of a fall of the carry bit signal *C 2 .

Therefore, in the half adder circuit 40 for the least significant digit, the carry-out bit *C 2 is generated by a NAND gate driver 41 to which the input bit A 1 and a fixed bit C 1 =‘1’ are provided. Further, in order to effectively use the output of the NAND gate 41 to constitute an exclusive-OR gate, a carry-in bit C 1 and an input bit A 1 are provided to an OR gate 42 , the outputs of the NAND gate 41 and the OR gate 42 are provided to a NAND gate 43 and the output thereof 43 is provided to an inverter 44 to take out an output bit D 1 . The operation of the half adder circuit 40 is represented by the following logic equation: D1 = ( C1 + A1 ) ·   *  ( C1 · A1 ) = ( C1 + A1 ) · (   *  C1 +   *  A1 ) = C1 ·   *  A1 +   *  C1 · A1 . 

Therefore, the half adder circuit 40 also functions as an exclusive-OR gate.

Although the edges of the carry bit signal *C 2 is blunted by passing through the transfer gates 212 to 214 , the waveform of the carry-out bit signal *C 2 is shaped by an inverter 324 for the most significant digit to obtain a carry-out bit signal C 5 with sharp edges.

Then, description will be given of a concrete comparison result between the circuits of FIGS. 2 and 11.

Although fifty gates are necessary if the incrementer of FIG. 11 is constituted of an ordinary circuit, only thirty-six gates are necessary to constitute the circuit of FIG. 2 . Further, the result of a simulation using virtual wiring capacitance in a 0.35 μm technology was such that a carry propagation time in the worst case wherein the input bits A 4 to A 1 are ‘1111’ was 3.8 ns for the circuit of FIG. 11, while being 2.6 ns for the circuit of FIG. 2 .

FIG. 3 shows a layout representation of the circuit of FIG. 2 which was used in the above simulation. In FIG. 3, wires are each shown by center lines, widths of metal wires are shown by cross-hatchings and widths of polysilicon gates by hatchings. Each PMOS transistor in an N well includes a polysilicon gate and P type diffused regions on both sides of this polysilicon gate. Each NMOS transistor in a P well includes a polysilicon gate and N type diffused regions on both sides of this polysilicon gate. Each interlayer contact is shown in the shape of a square.

According to the incrementer of the first example, not only can a circuit scale be down-sized, but an operation can also be faster.

FIG. 4 is a block diagram showing a 16 bit incrementer constructed of four 4 bit incrementers of FIG. 2 connected in cascade.

The bit C 1 is fixed at the power supply potential VDD and when 16 bit inputs A 1 to A 16 are provided, 17 bit outputs D 1 to D 16 and C 17 are asynchronously obtained by adding the C 1 =‘1’ to the 16 bit inputs A 1 to A 16 . Since the carries C 5 , C 8 , C 12 and C 17 outputted from the most significant digits of the respective 4 bit incrementers have enough driving abilities by respective inverters ( 324 ), reduction in operation speed due to blunting of signals is prevented from occurring.

Second Embodiment

FIG. 5 shows a half adder circuit of the second embodiment according to the present invention.

In this circuit, the bits *C 1 and *A 1 are provided to an exclusive-OR gate 11 to generate an output bit D 1 , wherein the carry-out bit *C 2 is not used for generation of the output bit D 1 . The other points are the same as those of the circuit of FIG. 1 .

FIG. 6 shows an example of FIG. 5 wherein the exclusive OR gate is of the same configuration as the circuit 11 A in FIG. 12 .

FIG. 7 is a circuit diagram showing a 4 bit incrementer to which the half adder circuit of FIG. 6 is applied. A half adder circuit 40 for the least significant digit is the same as that of FIG. 2 .

Since the operation of this circuit is apparent from the above description, description thereof is omitted.

Third Embodiment

FIG. 8 shows an incrementer of the third embodiment according to the present invention.

Since transfer gates 212 to 214 of binary carry logic circuits for the upper three digits are connected in series to each other similarly to FIG. 7, high speed carry propagation is achieved. A feature of this circuit is that carries to the upper digits are all non-inverted signals.

In a half adder circuit 40 A for the least significant digit, an AND gate 41 A is used instead of the NAND gate 41 of FIG. 2 .

In order to effectively use the output of this AND gate 41 A in constitution of an exclusive-OR gate, a fixed bit C 1 =‘1’ and an input bit A 1 are provided to a NOR gate 42 A, the outputs of the AND gate 41 A and the NOR gate 42 A are provided to a NOR gate 45 , and an output bit D 1 is taken out from the NOR gate 45 . The operation of the half adder circuit 40 A is represented by the following logic equation: D1 =      *  ( *  ( C1    + A1 ) + C1 · A1 ) = ( C1 + A1 ) · (   *  C1 +   *  A1 ) = C1 ·   *  A1 +   *  C1 · A1 . 

Therefore, the half adder circuit 40 A also functions as an exclusive-OR gate.

Each binary carry logic circuit 20 A for the other digits is different from the circuit 20 of FIG. 1 in that an NMOS transistor 23 A is connected between the data output of the transfer gate 212 and ground, and that the output of the inverter 22 is provided to the gate of the NMOS transistor 23 A.

When A 2 =‘1’, the transfer gate 212 and the NMOS transistor 23 A are on and off, respectively, to be C 3 =C 2 . When A 2 =‘0’, the transfer gate 212 and the NMOS transistor 23 A are off and on, respectively, to be C 3 =‘0’ regardless of the value of a carry-in bit C 2 from the lower digit. Therefore, C 3 =‘1’ only when A 2 =‘1’ and C 2 =‘1’, and C 3 =‘0’ in the other cases. That is, the binary carry logic circuit functions as an AND gate.

Features of the binary carry logic circuit 20 A in configuration are that on/off of the transfer gate 212 is reverse to that of the NMOS transistor 23 A, and that the logic value of the source potential, 0V, of the NMOS transistor 23 A is equal to that of the carry-in bit C 2 which is in an inactive state, which are the same as those of the above described circuit of FIG. 1 .

The carry-in bit C 2 and the input bit A 2 are provided to the exclusive-OR gate 11 A.

Further, the output data signal of the transfer gate 214 for the most significant digit becomes a carry-out bit signal CS whose waveform has been shaped through inverters 324 and 325 .

Since the operation of this circuit is apparent from the above description, description thereof is omitted.

Fourth Embodiment FIG. 9 shows a decrementer of the fourth embodiment according to the present invention.

Since in this circuit, transfer gates 212 to 214 of binary borrow logic circuits 20 X for the upper three digits are connected in series to each other, borrow propagation is achieved at a high speed.

In a half subtractor circuit 40 X for the least significant digit, a borrow-out bit *B 2 is generated by a NAND gate 41 and an inverter 46 with using an input bit A 1 and a fixed bit B 1 =‘1’, and *B 2 =‘0’ only when A 1 =‘0’ and B 1 =‘1’.

Further, in order to effectively use the outputs of the inverter 46 and the NAND gate 41 in constitution of an exclusive-OR gate, B 1 and the output *A 1 of the inverter 46 are provided to an OR gate 42 , the outputs of the NAND gate 41 and the OR gate 42 are provided to a NAND gate 43 and an output bit D 1 is taken out from the NAND gate 43 . The operation of the half subtractor circuit 40 X is represented by the following logic equation: D1 =      *  ( ( B1    +   *  A1 ) ·   *  ( B1    ·   *  A1 ) ) =   *  ( B1    +   *  A1 ) + ( B1    ·   *  A1 ) =   *  B1    ·   A1 + B1 ·   *  A1 .   

Therefore, the half subtractor circuit 40 X also functions as an exclusive-OR gate.

A binary borrow logic circuit 20 X has such connections that the on/off operation for an input bit A 2 is inverse to that of the binary carry logic circuit 20 of FIG. 2 .

When A 2 =‘0’, the transfer gate 212 and a PMOS transistor 23 are on and off, respectively, to be *B 3 =*B 2 . When A 2 =‘1’, the transfer gate 212 and the PMOS transistor 23 are off and on, respectively, to be *B 3 =‘1’ regardless of the value of the borrow-in bit *B 2 . Therefore, *B 3 =‘0’ only when A 2 =‘0’ and *B 2 =‘0’, and *B 3 =‘1’ in the other cases. That is, the following logic equation holds:

B 3 =*A 2 ·B 2 .  (3)

Features of the binary borrow logic circuit 20 X in configuration are that on/off of the transfer gate 212 is reverse to that of the PMOS transistor 23 , and that the logic value of the source potential, VDD, of the PMOS transistor 23 is equal to that of the borrow *B 2 which is in an inactive state, which are the same as those of the above described circuit of FIG. 1 .

In a logic circuit 30 A, a borrow B 2 obtained by inverting the borrow *B 2 with an inverter 37 and the output *A 2 of the inverter 22 are provided to an OR gate 38 , the output of the OR gate 38 and the borrow *B 3 are provided to a NAND gate 39 , and an output bit D 2 is taken out from the NAND gate 39 . The operation of the logic circuit 30 A is represented by the following logic equation: D2 =   *  ( ( B2    +   *  A2 ) ·   *  B3 ) =   *  B2    ·   A2 + B3 . 

The above equation is transformed to the following equation using the equation (3):

D 2 =A 2 ·*B 2 +*A 2 ·B 2 .  (4)

With such operation, a circuit for each digit functions as a half subtractor.

Since the operation of the decrementer of FIG. 9 is apparent from the above description, description thereof is omitted.

With the decrementer of the fourth example, the effects similar to the first embodiment are also obtained.

Fifth Embodiment

FIG. 10 shows a decrementer of the fifth embodiment according to the present invention.

Since in this circuit, transfer gates 212 to 214 of binary borrow logic circuits 20 Y for the upper three digits are connected in series to each other, borrow propagation is achieved at a high speed. A feature of this circuit is that a borrow-out bit to the upper side of each digit is a non- inverted signal.

In the half subtractor circuit 40 Y for the least significant digit, an AND gate 41 A is used instead of the NAND gate 41 of FIG. 9 .

In order to effectively use the output of the inverter 41 A in constitution of an exclusive-OR gate, a fixed bit B 1 =‘1’ and a signal *A 1 obtained by inverting an input bit A 1 with the inverter 46 are provided to an NOR gate 42 A, the outputs of the NAND gate 41 A and the NOR gate 42 A are provided to an OR gate 45 A, and an output bit D 1 is taken out from the OR gate 45 A. The operation of the half subtractor circuit 40 Y is represented by the following logic equation: D1 =      *  ( B1    +   *  A1 ) +   B1    ·   *  A1 =   *  B1    ·   A1 + B1 +   *  A1 .    

Therefore, the half subtractor circuit 40 Y also functions as an exclusive-OR gate.

In a binary borrow logic circuit 20 Y for any other digit, the configuration thereof is different from the binary borrow logic circuit 20 X of FIG. 9 in that an NMOS transistor 23 A is connected between the data output of the transfer gate 212 and ground, and that the input bit A 2 is provided to the gate of the NMOS transistor 23 A.

When A 2 =‘0’, the transfer gate 212 and the NMOS transistor 23 A are on and off, respectively, to be B 3 =B 2 . When A 2 =‘1’, the transfer gate 212 and the NMOS transistor 23 A are off and on, respectively, to be B 3 =‘0’ regardless of the value of the B 2 . Therefore, B 3 =‘1’ only when A 2 =‘0’ and B 2 =‘1’, and B 3 =‘0’ in the other cases. That is, the operation of the binary borrow logic circuit 20 Y is represented by the following logic equation:

B 3 =*A 2 ·B 2 .  (5)

Features of the binary borrow logic circuit 20 Y in configuration are that on/off of the transfer gate 212 is inverse to that of the NMOS transistor 23 A, and that the logic value of the source potential, 0V, of the NMOS transistor 23 A is equal to the logic value of the borrow B 2 which is in an inactive state, which are the same as those of the above described circuit of FIG. 1 .

The borrow B 2 and the input bit A 2 are provided to the exclusive-OR gate 11 A.

Further, the signal waveform on the data output of the transfer gate 214 for the most significant digit is shaped through inverters 324 and 325 to become a borrow B 5 .

Since the operation of the decrementer of FIG. 10 is apparent from the above description, description thereof is omitted.

Although preferred embodiments of the present invention has been described, it is to be understood that the invention is not limited thereto and that various changes and modifications may be made without departing from the spirit and scope of the invention.

For example, although in the above embodiments, description is given of a case where input bits and output bits are both of positive logic, the present invention also includes cases where one of the input bits or output bits is of negative logic and where both of the input and output bits are of negative logic. Further, a circuit for the least significant digit in an incrementer or a decrementer may be the same as that for an upper digit.