1. Field of the Invention
The present invention relates to a technique for carrying out a high-speed transmission of signals between LSIs or between devices formed by a plurality of LSIs. More particularly, the present invention relates to a multiplexer circuit that converts parallel data into serial data.
2. Description of the Related Art
Recently, the performance of components used in computers and other information processing apparatuses has been greatly improved. In particular, dramatic improvements have been made, for example, in the performance of processors and semiconductor memory devices such as SRAMs (Static Random Access Memories) and DRAMs (Dynamic Random Access Memories). The improvements in the performance of semiconductor memory devices, processors, and the like have come to the point where system performance cannot be improved further unless the speed of signal transmission between components or elements is increased.
Further, in recent years, along with the increase in the operation speed of LSIs, it has become necessary to provide a signal transmission system that can perform high-speed transmission of large-capacity signals between LSIs or between devices constructed of a plurality of LSIs. For example, in network infrastructures, high-speed transmission in the order of gigabits per second is required. As a result, a focus has been placed on a device called a “giga-bit SerDes (serializer and deserializer)”.
In the interface circuit that has the SerDes function, it is necessary to convert relatively low-speed parallel data received from a logic circuit that carries out a data processing like a network switching, into high-speed serial data at Gbps speed, and output the converted data, for example. As the data speed has become so fast, it has become necessary to improve time constants by preparing a separate current path, at a portion that has an upper limit in the operation frequency within the LSI circuit. Alternatively, it has become necessary to reduce a power source inductance portion, by using differential signals, in the data processing inside the LSI.
Conventionally, a multiplexer circuit that converts parallel data into serial data carries out a data processing by using multi-phase clock signals, as described in “DIGITAL SYSTEMS ENGINEERING”, Cambridge, 1998, by W, Dally et al. (for example, FIGS. 11–12 and FIGS. 11–25). In the conventional multiplexer circuits, it has been difficult to realize a high-speed operation.
The prior art and the problems associated with the prior art will be described in detail later with reference to accompanying drawings.
An object of the present invention is to provide a multiplexer circuit that can convert parallel data into serial data at high speed and synchronized with clock signals.
According to the present invention, there is provided a multiplexer circuit, having a plurality of multiplexer cells that receive parallel data, for converting the parallel data into serial data, and synchronized with a clock signal, wherein each of the multiplexer cells comprises a first load and a plurality of first conductivity type transistors that are connected in series between a first power source line and a second power source line; and a level-changing circuit that changes a connection node of adjacent first conductivity type transistors to a level of the first power source line.
The multiplexer circuit may take out an output of each of the multiplexer cells from a connection node of the first load and the first conductivity type transistors that are connected in series. A plurality of the level-changing circuits may be provided at different positions of the connection node of the adjacent first conductivity type transistors. The level-changing circuit may be formed by at least one second conductivity type transistor that is connected to the first power source line. The first load may be formed by a second conductivity type transistor.
The level-changing circuit may be formed by a second load that is connected to the first power source line and at least one first conductivity type transistor that is connected in series with the second load. The second load may be formed by a second conductivity type transistor. The first and second loads may be formed by second conductivity type transistors that are connected by cross coupling. The first and second loads may be formed by a parallel layout of second conductivity type transistors that are connected by cross coupling and second conductivity type transistors that are not connected by cross coupling. The first and second loads may be formed by second conductivity type transistors.
The level-changing circuit may maintain the symmetry of the level-changing circuit in layout. The level-changing circuit may be formed by two second conductivity type transistors that are connected to the first power source line. Each of the multiplexer cells may be formed as a differential circuit that processes differential signals. The multiplexer circuit may further comprise a latching circuit that is provided at the outputs of the plurality of multiplexer cells.
The first power source line may be a high-potential power source line, the second power source line is a low-potential power source line, and the first conductivity type transistor may be an nMOS transistor. The second conductivity type transistor may be a pMOS transistor. The multiplexer cells may be four multiplexer cells that are controlled based on four-phase clock signals.
The present invention will be more clearly understood from the description of the preferred embodiments as set forth below with reference to the accompanying drawings, wherein:
Before describing the embodiments of the present invention in detail, conventional multiplexer circuits and problems of these multiplexer circuits will be explained with reference to the drawings.
As shown in
In other words, as shown in
Clock signals /φ0 to /φ3 and data /d0 to /d3 denote inverted signals (signals of an opposite logic) of the clock signals φ0 to φ3 and the data d0 to d3, respectively. Therefore, the clock signal /φ0 is the same as the clock signal φ2, and the clock signal /φ1 is the same as the clock signal φ3, for example. A reference symbol Vc denotes a predetermined bias voltage that is applied to the gates of the transistors 1133 and 1143. Vr denotes a reference voltage that is applied to outputs OUT− and OUT+ via the loads 1144 and 1145 respectively.
The multiplexer circuit shown in
A differential amplifier that is constructed of the transistors 1131 to 1133 and the loads 1134 and 1135 acts as a pre-driver. Differential outputs p and /p of this pre-driver are supplied to differential inputs (the gates of the differential transistors 1141 and 1142) of an output-stage differential amplifier that is constructed of the transistors 1141 to 1143 and the loads 1144 and 1145. This output-stage differential amplifier outputs differential outputs OUT+ and OUT−.
The conventional multiplexer circuits shown in
First, according to the conventional multiplexer circuits explained with reference to
For example, assume that the parallel data d0 is at the high level “H”. When the clock signals φ3 and φ0 are both at the high level “H”, the transistors 1002 and 1003 are turned ON, and the potential of the output OUT is lowered to the low level “L” (Vss). Thereafter, when the clock signal φ3 has fallen from the high level “H” to the low level “L”, the transistor 1002 is turned OFF. At the same time, the clock signal φ1 rises from the low level “L” to the high level “H”, thereby to turn ON the transistor 1013. (At this time, the transistor 1011 is ON, as the clock signal φ0 is at the high level “H”.) Consequently, a level corresponding to the data d1 appears at the output OUT. When the parallel data d1 is at the low level “L” at this time, the output OUT changes to the high level “H” according to the data d1. In order to raise the output OUT to the high level “H” (Vdd) at high speed, it is necessary to make small the value of the resistance (pull-up resistance) RT. However, the lowering of the pull-up resistance RT brings about an increase in the power consumption. Therefore, in actual practice, it has been difficult to realize a high-speed operation in the circuit shown in
Next, according to the conventional multiplexer circuits explained with reference to
For example, the parallel data d0 and /d0 are supplied to the gates of the differential pair of transistors 1132 and 1131 of the pre-driver, via the transfer gates (transistors) (1110 and 1120) and (1111 and 1121) that are controlled based on the clock signals φ0 and /φ1, respectively. However, because of a delay in the two transfer gates that have been inserted in series, it has been difficult to realize a high-speed operation.
Embodiments of a multiplexer circuit relating to the present invention will be explained in detail below with reference to the accompanying drawings.
As shown in
As shown in
The load 101 is provided between a high-potential power source line (Vdd) and an output node (output Q). Transistors 102 to 104 are connected in series between this output Q and a low-potential power source line (Vss).
The transistor 102 receives an input data D (parallel data PD0 to PD3) at its gate, and drives the output Q. The transistors 103 and 104 carry out switching operation according to clock signals CLK0 and CLK1 that are supplied to their gates respectively. A connection node PN of the transistor 102 and the transistor 103 is connected to the high-potential power source line (Vdd).
In other words, the driving transistor 102 determines the output data (Q) according to the input data D, when the transistors 103 and 104 for switching (for data confirmation) that are controlled based on the clock signals CLK0 and CLK1 respectively are turned ON. The input data D are the parallel data PD0 to PD3 for the multiplexer cells 10 to 13 respectively. Further, the clock signals CLK0 to CLK1 are (φ0 and φ1), (φ1 and φ2), (φ2 and φ3), and (φ3 and φ0) for the multiplexer cells 10 to 13 respectively.
As shown in
As is clear from the comparison between
As explained above, based on the provision of pull-up positions at a plurality of locations, it becomes possible to shorten a transition time of each node (PN1 and PN2) to an initial status. As a result, it becomes possible to realize a higher-speed operation.
As shown in
The application of the multiplexer cell of the third embodiment to the multiplexer cell 10 shown in
As shown in
If the parallel data PD0 (D) is at the high level “H”, the transistor 113 is turned ON, the transistor 112 is turned OFF, and the clock signal φ0 (CLK1) changes from the low level “L” to the high level “H” (a period T0), based on the data PD0 at the high level “H”. Then, the transistor 114 is turned ON, the transistor 111 is turned OFF, and the node N1 is lowered to the low level “L” (the low-potential power source voltage Vss). Next, when the clock signal φ1 also changes from the low level “L” to the high level “H” (a period T1), the transistor 115 is turned ON, and the level of the node N1 becomes the output SD0.
On the other hand, if the parallel data PD0 (D) is at the low level “L”, the transistor 113 is turned OFF, the transistor 112 is turned ON, the node N1 is increased to the high level “H” (maintains the high-potential power source voltage Vdd), and the clock signal φ0 changes from the low level “L” to the high level “H” (the period T0), based on the data PD0 at the low level “L”. Then, the transistor 114 is turned ON, and the transistor 111 is turned OFF. However, the node N1 maintains the high level “H”. Next, when the clock signal φ1 also changes from the low level “L” to the high level “H” (the period T1), the transistor 115 is turned ON, and the level of the node N1 becomes the output SD0.
Further, when the clock signal φ0 changes from the high level “H” to the low level “L” (a period T2), the transistor 114 is turned OFF, the transistor 111 is turned ON, and the node N1 is pulled up to the high level “H” again. During this period T2, the clock signal φ1 is at the high level “H”, the transistor 115 is ON, and the level of the node N1 that has been pulled up to the high level “H” is transmitted to the output Q (SD1). However, when the multiplexer cell that operates next outputs the low level “L” (for example, when the parallel data PD1 at the high level “H” is supplied to the multiplexer cell 11 shown in
As explained above, according to the third embodiment, a NAND-type transistor carries out a selection based on the clock signal CLK0 (φ0), unlike the structure of the multiplexer circuit that has been explained with reference to
As shown in
As explained above, according to the fourth embodiment, the transistor 114 that carries out a selection based on the clock signal CLK0 is disposed at the low-potential power source line (Vss) side of the data input transistor 113. This is different from the structure of the multiplexer circuit explained with reference to
As shown in
The transistors 125 and 126 are provided between a differential amplifier section (the transistors 121 to 124) and a low-potential power source line (Vss). A clock CLK0 is supplied to the gate of the transistor 125, and a clock CLK1 is supplied to the gate of the transistor 126. The fifth embodiment corresponds to a structure of the first embodiment shown in
The application of the multiplexer cell of the fifth embodiment to the multiplexer cell 10 shown in
As explained above, according to the fifth embodiment, it is possible to realize a high-speed operation, by deleting the nMOS transistors (the transistors 1110, 1120, 1111, and 1121 in
As is clear from the comparison between
The multiplexer cell of the seventh embodiment includes both the transistors 121 and 122, and the transistors 127 and 128 of the fifth embodiment shown in
The multiplexer cell of the sixth embodiment shown in
The multiplexer cell of the seventh embodiment has the advantages of both the fifth embodiment and the sixth embodiment. The pMOS loads 121 and 122 pull up the output nodes (Q and QX) that are driven by the pair of differential transistors 123 and 124. At the same time, the cross-coupled loads 127 and 128 complementarily pull up these output nodes (Q and QX). With this arrangement, it is possible to reduce the drain load to less than that of the sixth embodiment, while keeping a complementary relationship. In other words, as the gates of the transistors 125 and 126, that are connected by cross coupling, are connected to the drains of the two transistors 122 and 126, and 121 and 125, that are connected in parallel respectively, it is possible to reduce the drain load. This contributes to the highspeed operation.
As is clear from the comparison between
A clock signal CLK0 that is supplied to the gate of an nMOS transistor 125 is supplied to the gates of the transistors 131 and 132 respectively. During a period while the clock signal CLK0 is at the low level “L”, and the transistor 125 is OFF, the transistors 131 and 132 are ON to pull up the node PN. The two pull-up transistors 131 and 132 are provided, in this case, in order to maintain symmetry of the circuit on the layout. Therefore, when it is possible to dispose a pull-up transistor at a center portion of a differential amplifier section (the transistors 121 to 124) on the layout, this pull-up transistor can be constituted by one transistor.
According to the multiplexer cell of the eighth embodiment, the transistors 131 and 132 are ON to pull up the node PN, during a period while the transistor 125 is OFF. With this arrangement, it becomes possible to shorten a transition time of the node PN to an initial status. As a result, it becomes possible to realize a high-speed operation.
As is clear from the comparison between
According to the multiplexer cell of the ninth embodiment, there are two pull-up positions (a plurality of positions) in a similar manner to that of the multiplexer cell of the second embodiment shown in
As is clear from a comparison between
The multiplexer cell of the eleventh embodiment is further provided with pMOS transistors 133 and 134 that pull up a connection node (PN2) of the transistors 125 and 126, in the multiplexer cell of the tenth embodiment shown in
According to the multiplexer cell of the eleventh embodiment, there are two pull-up positions (a plurality of positions) as in the multiplexer cell of the second embodiment shown in
The multiplexer cell of the twelfth embodiment includes both the transistors 121 and 122, and the transistors 127 and 128 of the eighth embodiment shown in
The pMOS loads 121 and 122 pull up the output nodes (Q and QX) that are driven by the pair of differential transistors 123 and 124. At the same time, the cross-coupled loads 127 and 128 complementarily pull up these output nodes (Q and QX). With this arrangement, it is possible to reduce the drain load more than that of the sixth embodiment, while keeping a complementary relationship.
In other words, as the gates of the transistors 125 and 126 that are connected by cross coupling are connected to the drains of the two transistors 122 and 126, and 121 and 125, that are connected in parallel respectively, it is possible to reduce the drain load. This contributes to the high-speed operation. The transistors 131 and 132 are ON to pull up the node PN, during a period while the transistor 125 is OFF. With this arrangement, it becomes possible to shorten a transition time of the node PN to an initial status. As a result, it becomes possible to realize a high-speed operation, in a similar manner to that of the eighth embodiment and the tenth embodiment respectively.
According to the multiplexer cell of the thirteenth embodiment, there are further provided pMOS transistors 133 and 134 that pull up a connection node (PN2) of the transistors 125 and 126, in the multiplexer cell of the twelfth embodiment shown in
According to the multiplexer cell of the thirteenth embodiment, there are two pull-up positions (a plurality of positions). This is similar to the multiplexer cell of the second embodiment shown in
According to the multiplexer circuit shown in
As explained above, according to the above embodiments of the present invention, a path that directly pulls up a node inside the multiplexer cell is introduced. Based on this arrangement, it is possible to improve the operation band of the multiplexer circuit. Further, a reduction in the number of stages of the nMOS transistors of the pull-up path is advantageous for high-speed operation, and it becomes possible to realize a high-speed multiplexer.
As explained above in detail, according to the present invention, it is possible to provide a multiplexer circuit that can convert parallel data into serial data at high speed and in synchronized with a clock signal.
Many different embodiments of the present invention may be constructed without departing from the spirit and scope of the present invention, and it should be understood that the present invention is not limited to the specific embodiments described in this specification, except as defined in the appended claims.
Number | Date | Country | Kind |
---|---|---|---|
2001-322375 | Oct 2001 | JP | national |
Number | Name | Date | Kind |
---|---|---|---|
4932027 | Scharrer | Jun 1990 | A |
5654660 | Orgill et al. | Aug 1997 | A |
5761208 | Muramatsu | Jun 1998 | A |
6188339 | Hasegawa | Feb 2001 | B1 |
6208193 | Moyal et al. | Mar 2001 | B1 |
6310509 | Davenport | Oct 2001 | B1 |
20010007577 | Measor | Jul 2001 | A1 |
Number | Date | Country |
---|---|---|
1 079 300 | Feb 2001 | EP |
01-309512 | Dec 1989 | JP |
05-191239 | Jul 1993 | JP |
09-116405 | May 1997 | JP |
2000-236244 | Aug 2000 | JP |
Number | Date | Country | |
---|---|---|---|
20030076821 A1 | Apr 2003 | US |