US20230115318A1

US20230115318A1 - Apparatus and methods for high-speed drivers

Info

Publication number: US20230115318A1
Application number: US17/450,585
Authority: US
Inventors: Wei-Hung Chen
Original assignee: Analog Devices Inc
Current assignee: Analog Devices Inc
Priority date: 2021-10-12
Filing date: 2021-10-12
Publication date: 2023-04-13
Anticipated expiration: 2041-10-12
Also published as: US11640367B1; US12072832B2; US20230229620A1

Abstract

Apparatus and methods for high-speed drivers are provided herein. In certain embodiments, a high-speed driver multiplexes two or more data streams. The high-speed driver is implemented with a mux-then-driver topology that provides multiplexing in a predriver circuit. Thus, the multiplexer is eliminated from the full rate output path to relax timing. Driver amplitude control schemes are also disclosed in which a controllable driver includes a group of differential series source transistor (SST) driver slices that are connected in parallel with one another to drive a pair of output terminals, and a group of attenuator slices that are connected in parallel with one another across the pair of output terminals. Additionally, the controllable driver includes a control circuit that activates an attenuator slice for each SST driver slice that is decommissioned to provide output amplitude control.

Description

FIELD OF THE DISCLOSURE

Embodiments of the invention relate to electronics, and more particularly, to driver circuits for high-speed communications.

BACKGROUND

Serializer/deserializer (SerDes) systems can be used in a variety of applications such as telecommunications, optical networks, and/or chip-to-chip communication.
A SerDes system includes a serializer that converts two or more parallel input data streams into a high-speed serial data stream, and a deserializer that converts the high-speed serial data stream into two or more parallel output data streams of reduced speed. Thus, data transmission can be provided over a reduced number of lines to lower pin count.

SUMMARY OF THE DISCLOSURE

Apparatus and methods for high-speed drivers are provided herein. In certain embodiments, a high-speed driver multiplexes two or more data streams. The high-speed driver is implemented with a mux-then-driver topology that provides multiplexing in a predriver circuit. Thus, the multiplexer is eliminated from the full rate output path to relax timing. Driver amplitude control schemes are also disclosed in which a controllable driver includes a group of differential series source transistor (SST) driver slices that are connected in parallel with one another to drive a pair of output terminals, and a group of attenuator slices that are connected in parallel with one another across the pair of output terminals. Additionally, the controllable driver includes a control circuit that activates an attenuator slice for each SST driver slice that is decommissioned to provide output amplitude control.
In one aspect, a driver circuit includes an output terminal configured to provide an output data stream, and a first driver subcircuit configured to receive a first input data stream of a lower bit rate than the output data stream, and to drive the output terminal based on the first input data stream in response to a transition of a first clock signal. The first driver subcircuit includes a first driver transistor connected between a high supply voltage and the output terminal with no other transistors therebetween, a second driver transistor connected between a low supply voltage and the output terminal with no other transistors therebetween, a first pull-up predriver circuit configured to control the first driver transistor, and a first pull-down predriver configured to control the second driver transistor.
In another aspect, a serializer/deserializer (SerDes) system includes a deserializer, and a serializer including a driver. The driver includes an output terminal configured to provide an output data stream to the deserializer, and a first driver subcircuit configured to receive a first input data stream of a lower bit rate than the output data stream, and to drive the output terminal based on the first input data stream in response to a transition of a first clock signal. The first driver subcircuit includes a first driver transistor connected between a high supply voltage and the output terminal with no other transistors therebetween, a second driver transistor connected between a low supply voltage and the output terminal with no other transistors therebetween, a first pull-up predriver circuit configured to control the first driver transistor, and a first pull-down predriver configured to control the second driver transistor.
In another aspect, a method of multiplexing data streams is provided. The method includes providing an output data stream on an output terminal, receiving a first input data stream of a lower bit rate than the output data stream as an input to a first driver subcircuit, and driving the output terminal based on the first input data stream in response to a transition of a first clock signal using the first driver subcircuit, including controlling a first driver transistor connected between a high supply voltage and the output terminal with no other transistors therebetween using a first pull-up predriver, and controlling a second driver transistor connected between a low supply voltage and the output terminal with no other transistors therebetween using a first pull-down predriver.
In another aspect, a driver circuit includes a pair of output terminals configured to provide a differential output signal, a plurality of differential series source transistor (SST) driver slices electrically connected in parallel with one another and configured to drive the pair of output terminals, a plurality of attenuator slices connected in parallel with one another across the pair of output terminals, and a control circuit configured to selectively deactivate one or more of the differential SST driver slices to control an amplitude of the differential output signal, and to enable a corresponding number of the attenuator slices to provide output impedance compensation.
In another aspect, a method of output swing control in a driver circuit is provided. The method includes providing a differential output signal on a pair of output terminals, driving the pair of output terminals using a plurality of differential series source transistor (SST) driver slices electrically connected in parallel with one another, deactivating one or more of the differential SST driver slices to control an amplitude of the differential output signal, and enabling a corresponding number of a plurality of attenuator slices to provide output impedance compensation, wherein the plurality of attenuator slices are connected in parallel with one another across the pair of output terminals.
In another aspect, a serializer/deserializer (SerDes) system includes a deserializer, and a serializer comprising a driver including a pair of output terminals configured to provide a differential output signal to the deserializer, a plurality of differential series source transistor (SST) driver slices electrically connected in parallel with one another and configured to drive the pair of output terminals, a plurality of attenuator slices connected in parallel with one another across the pair of output terminals, and a control circuit configured to selectively deactivate one or more of the differential SST driver slices to control an amplitude of the differential output signal, and to enable a corresponding number of the attenuator slices to provide output impedance compensation.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1A is a schematic diagram of one embodiment of a multiplexing driver.

FIG. 1B is one example of a timing diagram for the multiplexing driver of FIG. 1A.

FIG. 2A is a schematic diagram of another embodiment of a multiplexing driver.

FIG. 2B is a schematic diagram of another embodiment of a multiplexing driver.

FIG. 3A is a schematic diagram of one embodiment of a driver half circuit for a multiplexing driver.

FIG. 3B is one example of a timing diagram for the driver half circuit of FIG. 3A.

FIG. 4A is a schematic diagram of another embodiment of a multiplexing driver.

FIG. 4B is one example of a timing diagram for the multiplexing driver of FIG. 4A.

FIG. 5A is a schematic diagram of another embodiment of a multiplexing driver.

FIG. 5B is one example of a timing diagram for the multiplexing driver of FIG. 5A.

FIG. 6 is a schematic diagram of another embodiment of a multiplexing driver.

FIG. 7A is a schematic diagram of one embodiment of a pull-up predriver circuit for a multiplexing driver.

FIG. 7B is one example of a timing diagram for the pull-up predriver circuit of FIG. 7A.

FIG. 8A is a schematic diagram of one embodiment of a driver quarter circuit for a multiplexing driver.

FIG. 8B is one example of a timing diagram for the driver quarter circuit of FIG. 8A.

FIG. 9 is a schematic diagram of one embodiment of a driver with controllable swing and constant output impedance.

FIG. 10A is a schematic diagram of another embodiment of a driver with controllable swing and constant output impedance.

FIG. 10B is a circuit diagram of a portion of the driver of FIG. 10A.

FIG. 11 is a graph of one example of driver amplitude reduction versus attenuation setting for the driver of FIGS. 10A and 10B.

FIG. 12A is a graph of one example comparison of driver output swing versus attenuation setting for two implementations of drivers.

FIG. 12B is a graph of one example comparison of driver current versus attenuation setting for two implementations of drivers.

FIG. 12C is a graph of one example comparison of driver output impedance versus attenuation setting for two implementations of drivers.

FIG. 13 is a schematic diagram of one embodiment of a SerDes system.

DETAILED DESCRIPTION OF EMBODIMENTS

The following detailed description of embodiments presents various descriptions of specific embodiments of the invention. However, the invention can be embodied in a multitude of different ways. In this description, reference is made to the drawings where like reference numerals may indicate identical or functionally similar elements. It will be understood that elements illustrated in the figures are not necessarily drawn to scale. Moreover, it will be understood that certain embodiments can include more elements than illustrated in a drawing and/or a subset of the elements illustrated in a drawing. Further, some embodiments can incorporate any suitable combination of features from two or more drawings.
A SerDes system can include a serializer that generates a high-speed serial data stream based on data streams associated with different time offsets or delays. For instance, in a half rate SerDes system, a serializer can combine a first half rate data stream and a second half rate data stream into a full rate data stream having a bit period, with the second half rate data stream delayed by the bit period relative to the first half rate data stream.
In certain embodiments herein, high-speed drivers for multiplexing two or more data streams are provided. The high-speed driver is implemented with a mux-then-driver topology that provides multiplexing in a predriver circuit. Thus, the multiplexer is eliminated from the full rate output path to relax timing. For example, in an implementation with two data streams, a timing constraint is relaxed by a factor of two.
Moreover, implementing the multiplexer in the predriver reduces a size of the multiplexing transistors relative to an implementation in which an explicit T-gate multiplexer is included along the output signal path. By reducing the size of the multiplexing transistors, the multiplexer size is shrunk and the total capacitance is reduced to achieve reduced power and a tighter output eye diagram.
FIG. 1A is a schematic diagram of one embodiment of a multiplexing driver 10. FIG. 1B is one example of a timing diagram for the multiplexing driver 10 of FIG. 1A.
With reference to FIG. 1A, the multiplexing driver 10 operates to multiplex an odd half rate data stream D_ODD(also referred to herein as an odd data stream) and an even half rate data stream D_EVEN(also referred to herein as an even data stream) to generate a full rate data stream D_OUT. Thus, the multiplexing driver 10 provides two-way interleaving.
The multiplexing driver 10 receives the odd data stream D_ODDand the even data stream D_EVEN, as well as a first clock signal CK₀and a second clock signal CK₁₈₀used for controlling timing of the data streams. In the example of FIG. 1B, the value of the odd data stream D_ODDbecomes ready at the input of the multiplexing driver 10 before a rising edge of the first clock signal CK₀, and the value is transferred to the full rate data stream D_OUTin response to a rising edge of the second clock signal CK₁₈₀. Additionally, the value of the even data stream D_EVENbecomes ready at the input of the multiplexing driver 10 before a rising edge of the second clock signal CK₁₈₀, and the value is transferred to the full rate data stream D_OUTin response to a rising edge of the first clock signal CK₀.
In the example of FIG. 1B, annotations for an example full data rate of 32 gigabits per second (Gbps) are depicted. In this example, the half rate data streams operate at 16 Gbps and a corresponding bit interval of 62.5 picoseconds (ps), while the full rate data stream has a 31.25 ps bit interval. The first clock signal CK₀and the second clock signal CK₁₈₀operate at 16 gigahertz (GHz) with a bit period offset from one another.
When operating at a full data rate of 32 Gbps, the 31.25 ps bit period approaches the process limit for certain processes, such as 16 nanometer (nm) processes associated with an inverter fan-out two delay of about 5 ps.
To implement the multiplexing driver 10, the odd data stream D_ODDcould be provided to a first inverter that drives a first T-gate multiplexer, and the even data stream D_EVENcould be provided to a second inverter that drives a second T-gate multiplexer. Additionally, the first T-gate multiplexer could pass the odd data stream D_ODDto the output based on timing of the second clock signal CLK₁₈₀(for instance, in response to a rising edge), while the second T-gate multiplexer could pass the even data stream D_EVENto the output based on timing of the first clock signal CLK₀(for instance, in response to a rising edge).
However, such an implementation has high output resistance due to a T-gate multiplexer being in series with each driver inverter. Furthermore, a large T-gate multiplexer size (to keep the output resistance low due to the series combination of transistors) results in high parasitic capacitance. Moreover, including a series output resistor for impedance matching (for instance, 50 Ohm) can further raise output resistance and slow timing. Furthermore, asymmetries in the logic gates used to drive the T-gate multiplexers leads to imbalances in rise/fall times and a data eye that is bimodal.
In certain embodiments herein, the multiplexing driver 10 is implemented using a mux-then-driver topology that provides multiplexing in a predriver circuit. Thus, the multiplexer is eliminated from the full rate output path, thereby relaxing timing constraints to allow handling of data streams of higher bit rates.
FIG. 2A is a schematic diagram of another embodiment of a multiplexing driver 40. The multiplexing driver 40 includes a first driver half circuit 11, a second driver half circuit 12, an output resistor 13, and an output pin or pad 14.
As shown in FIG. 2A, the first driver half circuit 11 receives the even data stream D_EVEN, the first clock signal CK₀, and the second clock signal CK₁₈₀, while the second driver half circuit 12 receives the odd data stream D_ODD, the first clock signal CK₀, and the second clock signal CK₁₈₀. The first driver half circuit 11 operates to provide the even data stream D_EVENto the output pad 14 based on timing of the first clock signal CK₀(for instance, in response to a rising edge), while the second driver half circuit 12 operates to provide the odd data stream D_ODDto the output pad 14 based on timing of the second clock signal CK₁₈₀(for instance, in response to a rising edge).
Thus, the first driver half circuit 11 and the second driver half circuit 12 operate in an alternating or ping-pong sequence.
The first driver half circuit 11 and the second driver half circuit 12 each include an output connected to the output pad 14 by way of the output resistor 13. Including the output resistor 13 can aid in achieving a desired output impedance match, for instance, 50 Ohms or other desired output impedance value. The output resistor 13 can be implemented in a wide variety of ways including, but not limited to, using polysilicon or other resistive material having a geometry selected to achieve a target resistance value. In certain implementations, the output resistor 13 is trimmable and/or otherwise controllable to achieve an output resistance that provides compensation for variation.
In the illustrated embodiment, the first driver half circuit 11 includes a first driver p-type field effect transistor (PFET) 21, a first driver n-type field effect transistor (NFET) 22, a first pull-up predriver circuit 23, and a first pull-down predriver circuit 24. Additionally, the second driver half circuit 12 includes a second driver PFET 31, a second driver NFET 32, a second pull-up predriver circuit 33, and a second pull-down predriver circuit 34. The predriver circuits are implemented with multiplexing in accordance with the teachings herein.
Accordingly, the multiplexing driver 40 is advantageously implemented with multiplexing that is implemented in predriver stages, thereby allowing a connection to the output pad 14 that goes through a low number of transistors.
For example, as shown in FIG. 2A, the first driver half circuit 11 can drive the output pad 14 logically high using the first driver PFET 21, which is connected in series with the output resistor 13 between a power supply voltage V_DD(also referred to herein as a high supply voltage) and the output pad 14 with no other transistors therebetween. Likewise, the first driver half circuit 11 can drive the output pad 14 logically low using the first driver NFET 22, which is connected in series with the output resistor 13 between a ground voltage (also referred to herein as a low supply voltage or V_SS) and the output pad 14 with no other transistors therebetween.
Accordingly, timing constraints are relaxed by eliminating a multiplexer (for instance, a T-gate multiplexer in cascade with a driver inverter) from the output resistance path.
The NFETs and PFETs can be implemented in a wide variety of ways. In one example, the multiplexing driver 40 is fabricated in a complementary metal oxide semiconductor (CMOS) process, and the NFETs correspond to n-type metal oxide semiconductor (NMOS) transistors while the PFETs correspond to p-type metal oxide semiconductor (PMOS) transistors.
FIG. 2B is a schematic diagram of another embodiment of a multiplexing driver 50. The multiplexing driver 50 of FIG. 2B is similar to the multiplexing driver 40 of FIG. 2A, except that the multiplexing driver 50 is implemented in a differential configuration.
For example, the multiplexing driver 50 includes a first non-inverted (+) driver half circuit 11 a, a second non-inverted driver half circuit 12 a, a first output resistor 13 a, and a first output pad 14 a. The first non-inverted driver half circuit 11 a operates to provide the non-inverted even data stream D_EVEN+ to the first output pad 14 a (which provides D_OUT+) based on timing of the first non-inverted clock signal CK₀₊, while the second non-inverted driver half circuit 12 a operates to provide the non-inverted odd data stream D_ODD+ to the first output pad 14 a based on timing of the second non-inverted clock signal CK₁₈₀₊. The first non-inverted driver half circuit 11 a includes a first driver PFET 21 a, a first driver NFET 22 a, a first pull-up predriver circuit 23 a, and a first pull-down predriver circuit 24 a, while the second non-inverted driver half circuit 12 a includes a second driver PFET 31 a, a second driver NFET 32 a, a second pull-up predriver circuit 33 a, and a second pull-down predriver circuit 34 a.
With continuing reference to FIG. 2B, the first inverted driver half circuit 11 b operates to provide the inverted even data stream D_EVEN− to the second output pad 14 b (which provides D_OUT−) based on timing of the first inverted clock signal CK₀₋, while the second inverted driver half circuit 12 b operates to provide the inverted odd data stream D_ODD− to the second output pad 14 b based on timing of the second inverted clock signal CK₁₈₀₋. The first inverted driver half circuit 11 b includes a first driver PFET 21 b, a first driver NFET 22 b, a first pull-up predriver circuit 23 b, and a first pull-down predriver circuit 24 b, while the second inverted driver half circuit 12 b includes a second driver PFET 31 b, a second driver NFET 32 b, a second pull-up predriver circuit 33 b, and a second pull-down predriver circuit 34 b.
Any of the driver circuits herein can be implemented differentially. By implementing a multiplexing predriver differentially, enhanced immunity against common-mode noise can be achieved.
FIG. 3A is a schematic diagram of one embodiment of a driver half circuit 70 for a multiplexing driver. For example, the driver half circuit 70 can connect to an output resistor 13 and an output pad 14 of the multiplexing driver as shown. FIG. 3B is one example of a timing diagram for the driver half circuit 70 of FIG. 3A when operating at 32 Gbps.
As shown in FIG. 3A, the driver half circuit 70 receives the even data stream D_EVEN, the first clock signal CK₀, and the second clock signal CK₁₈₀. The driver half circuit 70 includes a driver PFET 21 (also referred to as driver PFET M3), a driver NFET 22 (also referred to as driver NFET M3′), a pull-up predriver circuit 63, and a pull-down predriver circuit 64.
In the illustrated embodiment, the pull-up predriver circuit 63 includes a pull-down data NFET M1, a pull-up data PFET M2, a multiplexing NFET M4, and a pre-charge PFET M5. As shown in FIG. 3A, the pull-up predriver circuit 63 controls activation of the driver PFET M3 at a node Y that is pre-charged to V_DDby the pre-charge PFET M5 when the first clock signal CK₀is low. Additionally, the pull-up data PFET M2 and the pull-down data NFET M1 control the node X to one of V_DDor ground (V_SS) based on a state of the even data stream D_EVEN. In response to the first clock signal CK₀going high, the multiplexing NFET M4 passes the value of node X to node Y to thereby control the driver PFET M3.
With continuing reference to FIG. 3A, the pull-down predriver circuit 64 includes a pull-up data PFET M1′, a pull-down data NFET M2′, a multiplexing PFET M4′, and a pre-charge NFET M5′. As shown in FIG. 3A, the pull-down predriver circuit 64 controls activation of the driver NFET M3′ at a node Y′ that is pre-charged to ground by the pre-charge NFET M5′ when the second clock signal CK₁₈₀is high. Additionally, the pull-up data PFET M1′ and the pull-down data NFET M2′ control the node X′ to one of V_DDor ground based on a state of the even data stream D_EVEN. In response to the second clock signal CK₁₈₀going low, the multiplexing PFET M4′ passes the value of node X′ to node Y′ to thereby control the driver MFET M3′.
Thus, the pull-up predriver circuit 63 and the pull-down predriver circuit 64 operate in a first phase associated with pre-charge followed by a second phase in which the output is pulled up or down based on a state of the even data stream D_EVEN.
By implementing the driver half circuit 70 in this manner, a number of performance enhancements are achieved including, but not limited to, a relaxed setup time (t_setup).
Table 1 below provides a summary of operation of the pull-up predriver circuit 63 over the first phase and the second phase.

TABLE 1

Phase	CK₀	CK₁₈₀	D_EVEN	node Y	M4	M5

1	Low	high	settling	pre-charge	off	on
				to V _DD
2	High	low	transparent	pull-down	on	off
				when
				D_EVEN= 1

FIG. 4A is a schematic diagram of another embodiment of a multiplexing driver 100. FIG. 4B is one example of a timing diagram for the multiplexing driver 100 of FIG. 4A when operating at 32 Gbps. The multiplexing driver 100 includes a first driver half circuit 70, a second driver half circuit 80, an output resistor 13, and an output pad 14.
The multiplexing driver 100 includes two half driver circuits implemented in accordance with the embodiment of FIG. 3A. For example, the multiplexing driver 100 includes the first half driver circuit 70, as described earlier with respect to FIG. 3A. The multiplexing driver 100 further includes the second half driver circuit 80 used to control the output data stream D_OUTbased on the odd data stream D_ODDand timing of the first clock signal CK₀and the second clock signal CK₁₈₀.
As shown in FIG. 4A, the second half driver circuit 80 includes a pull-down data NFET M1″ (for pulling down node X″ when D_ODDis high), a pull-up data PFET M2″, a driver PFET M3″, a multiplexing NFET M4″ (controlled by CK₁₈₀), a pre-charge PFET M5″ (for pre-charging node Y″ to V_DDwhen CK₁₈₀is low), a pull-up data PFET M1″′ (for pulling up node X″′ when D_ODDis low), a pull-down data NFET M2″′, a multiplexing PFET M4″′ (controlled by CK₀), and a pre-charge NFET M5″′ (for pre-charging node Y″′ to ground when CK₀is high).
FIG. 5A is a schematic diagram of another embodiment of a multiplexing driver 210. FIG. 5B is one example of a timing diagram for the multiplexing driver 210 of FIG. 5A.
With reference to FIG. 5A, the multiplexing driver 210 generates a full rate data stream D_OUTby multiplexing a first quarter rate data stream D₀, a second quarter rate data stream D₉₀, a third quarter rate data stream D₁₈₀, and a fourth quarter rate data stream D₂₇₀. Thus, the multiplexing driver 210 provides four-way interleaving.
In addition to receiving the quarter rate data streams, the multiplexing driver 210 receives a first clock signal CK₀, a second clock signal CK₉₀, a third clock signal CK₁₈₀, and a fourth clock signal CK₂₇₀that are offset in phase from one another (by a bit interval of D_OUT).
In the example of FIG. 5B, annotations for operation at a full data rate of 56 Gbps are depicted. In this example, the quarter rate data streams operate at 14 Gbps and a corresponding bit interval of 71.8 ps, while the full rate data stream has a 17.8 ps bit interval. Furthermore, the clock signals operate at a 14 GHz with a bit period offset from one another.
By providing 4-way interleaving, higher output data rate can be achieved relative to 2-way interleaving or no interleaving.
In certain embodiments herein, the multiplexing driver 210 is implemented using a mux-then-driver topology that provides multiplexing in a predriver circuit. Thus, the multiplexer is eliminated from the full rate output path to relax timing constraints.
FIG. 6 is a schematic diagram of another embodiment of a multiplexing driver 260. The multiplexing driver 260 includes a first driver quarter circuit 211, a second driver quarter circuit 212, a third driver quarter circuit 213, a fourth driver quarter circuit 214, an output resistor 215, and an output pin or pad 216.
As shown in FIG. 6 , the first driver quarter circuit 211 receives the first data stream D₀, the second driver quarter circuit 212 receives the second data stream D₉₀, the third driver quarter circuit 213 receives the third data stream D₁₈₀, and the fourth driver quarter circuit 214 receives the fourth data stream D₂₇₀. The driver quarter circuits 211-214 are interleaved to drive the output pad 216 with their respective data streams based on timing of the first clock signal CK₀, the second clock signal CK₉₀, the third clock signal CK₁₈₀, and the fourth clock signal CK₂₇₀.
The driver quarter circuits 211-214 each include an output connected to the output pad 216 by way of the output resistor 215. Including the output resistor 215 can aid in achieving a desired output impedance match, for instance, 50 Ohms or other desired output impedance value.
In the illustrated embodiment, the first driver quarter circuit 211 includes a first driver PFET 221, a first driver NFET 222, a first pull-up predriver circuit 223 for controlling the first driver PFET 221, and a first pull-down predriver circuit 224 for controlling the first driver NFET 222. Additionally, the second driver quarter circuit 212 includes a second driver PFET 231, a second driver NFET 232, a second pull-up predriver circuit 233 for controlling the second driver PFET 231, and a second pull-down predriver circuit 234 for controlling the second driver NFET 232. Furthermore, the third driver quarter circuit 213 includes a third driver PFET 241, a third driver NFET 242, a third pull-up predriver circuit 243 for controlling the third driver PFET 241, and a third pull-down predriver circuit 244 for controlling the third driver NFET 242. Additionally, the fourth driver quarter circuit 214 includes a fourth driver PFET 251, a fourth driver NFET 252, a fourth pull-up predriver circuit 253 for controlling the fourth driver PFET 251, and a fourth pull-down predriver circuit 254 for controlling the fourth driver NFET 252.
The multiplexing driver 260 is advantageously implemented with multiplexing in predriver stages, thereby allowing a connection to the output pad 216 that goes through a low number of transistors. For example, as shown in FIG. 6 , each of the driver quarter circuits 211-214 can drive the output pad 216 to V_DDor ground (V_SS) through a single transistor. Accordingly, timing constraints are relaxed by eliminating a multiplexer (for instance, a T-gate multiplexer in cascade with a driver inverter) from the output resistance path.
FIG. 7A is a schematic diagram of one embodiment of a pull-up predriver circuit 280 for a multiplexing driver. FIG. 7B is one example of a timing diagram for the pull-up predriver circuit 280 of FIG. 7A.
The pull-up predriver circuit 280 of FIG. 7A illustrates one embodiment of the pull-up predriver circuit 233 of FIG. 6 , and thus the pull-up predriver circuit 280 can be included as part of a driver quarter circuit. As shown in FIG. 7A, connections of the pull-up predriver circuit 280 to the second driver PFET 231 (also referred to as driver PFET M3), the output resistor 215, and the output pad 216 are depicted.
In the illustrated embodiment, the pull-up predriver circuit 280 includes a first multiplexing NFET M1, a data NFET M2, a second multiplexing NFET M4, a first pre-charge PFET M5, and a second pre-charge PFET M6.
As shown in FIG. 7A, the pull-up predriver circuit 280 controls activation of the driver PFET M3 at a node Y that is pre-charged to V_DDby the first pre-charge PFET M5 when the first clock signal CK₀is low and by the second pre-charge PFET M6 when the second clock signal CK₉₀is low. The node Z is pulled low in response to the second clock signal CK90 going high, and the data NFET M2 pulls the node X to ground when the data stream D₉₀is also high. The value of node X is passed to node Y through the second multiplexing NFET M4, which is controlled by the first clock signal CK₀.
Thus, the pull-up predriver circuit 280 operates in multiple circuit phases. Table 2 below provides a summary of operation of the pull-up predriver circuit 280 over the phases.

TABLE 2

Phase	CK₀	CK₉₀	D₉₀	node Y	M4	M1

1	Low	low	settling	pre-charge	off	off
			after	to V_DD
			transition	(M5 + M6)
2	High	low	finishing	pre-charge	on	off
			settling	to V_DD
				(M6)
3	high	high	no change	pull-down	on	on
				if D90 = 1
4	Low	high	no change	pre-charge	off	on
				to V_DD
				(M5)

The multi-phase circuit sequence is also graphically depicted in FIG. 7B.
The pull-up predriver circuit 280 of FIG. 7A advantageously includes the fourth phase to turn-off the data NFET M4 and pre-charge node Y quickly (for example, as quickly as CK0 drops below the threshold voltage of the first pre-charge PFET M5) rather than waiting for the whole transistor sequence M1, M2, and M4 to turn off (in order).
Moreover, this pre-charge scheme provides fast performance time and/or low parasitic capacitance while avoiding a need for a separate pre-charge transistor for node X (for instance an additional transistor directly connected between V_DDand node X).
FIG. 8A is a schematic diagram of one embodiment of a driver quarter circuit 300 for a multiplexing driver. FIG. 8B is one example of a timing diagram for the driver quarter circuit 300 of FIG. 8A for operation at 56 Gbps.
The driver quarter circuit 300 is depicted with connections to the output resistor 215 and the output pad 216. The driver quarter circuit 300 includes a driver PFET 231 (also referred to as driver PFET M3), a driver NFET 232 (also referred to as driver NFET M3′), a pull-up predriver circuit 280, and a pull-down predriver circuit 290.
The driver quarter circuit 300 is implemented with predriver circuits implemented in accordance with the embodiment of FIG. 7A. For example, the driver quarter circuit 300 includes the pull-up predriver circuit 280 as discussed above with reference to FIG. 7A, as well as the pull-down predriver circuit 290 which corresponds to a complementary version of the pull-up predriver circuit 280 in which transistor polarities and power supply connections are reversed and the clock signals are delayed by 180 degrees to provide inversion.
As shown in FIG. 8A, the pull-down predriver circuit 290 includes a first multiplexing PFET M1′ (controlled by CK₂₇₀and connected between V_DDand node Z′), a data PFET M2′ (controlled by D₉₀and connected between node Z′ and node X′), a second multiplexing PFET M4′ (controlled by CK₁₈₀and connected between node X′ and node Y′), a first pre-charge NFET M5′ (controlled by CK₁₈₀and connected between node Y′ and ground), and a second pre-charge NFET M6′ (controlled by CK₂₇₀and connected between node Y′ and ground).
The multi-phase circuit sequence of the driver quarter circuit 300 of FIG. 8A is graphically depicted in FIG. 8B.
Drivers with Controllable Output Swing and Constant Output Impedance
In certain applications, such as SerDes, it is desirable for an output driver to have constant output impedance while at the same time having controllable output swing to achieve desired signal amplitude. For example, implementing a driver with variable output amplitude control allows for enhanced flexibility for achieving desired signal level. However, it is desirable for the change in output amplitude or swing to not degrade performance by changing the output impedance from a desired level.
Although current mode logic (CML) drivers can realize controllable swing and constant output impedance, CML drivers suffer from a number of undesirable characteristics, such as high power consumption. Series source transistor (SST) drivers offer improved power performance, but suffer from varying output impedance when the driver transistor size is changed to adjust output amplitude.
SST drivers with controllable output swing and constant output impedance are provided. In certain embodiments herein, a controllable driver includes a group of differential SST driver slices that are connected in parallel with one another to drive a pair of output terminals providing a differential output signal, and a group of attenuator slices that are connected in parallel with one another across the pair of output terminals. Each attenuator slice can be implemented to have an on-state resistance about equal to an on-state resistance of one of the differential SST driver slices. Additionally, the controllable driver includes a control circuit that activates an attenuator slice for each SST driver slice that is decommissioned to provide amplitude control. Thus, for every differential SST driver slice that is disabled for amplitude control, an attenuator slice is enabled.
Thus, the combined total number of active SST driver slices and active attenuator slices remains constant, and the output impedance remains at a desired value (for instance, 50 Ohms).
In certain implementations, the control circuit can be implemented to also disable any clock and data path circuits used to drive a differential SST driver slice that is disabled for amplitude control. By implementing the controllable driver in this manner, any additional current draw by the attenuator array is outweighed by a current reduction arising from disabling the clock and data path circuits leading to the decommissioned slices.
FIG. 9 is a schematic diagram of one embodiment of a driver 400 with controllable swing and constant output impedance. The driver 400 includes differential SST slices 401 a, 401 b, . . . 401 i operating in parallel with one another to drive a pair of output terminals V_OUT+/V_OUT−. The driver 400 further includes attenuator slices 402 a, 402 b, . . . 402 j in parallel with one another across the pair of differential output terminals V_OUT+, V_OUT−. The driver 400 further includes a control circuit 403 and data/clock path slices 404 a, 404 b, . . . 404 i.
In the illustrated embodiment, the control circuit 403 generates a first group of enable signals EN1 a, EN1 b, . . . EN1 i for enabling the SST slices 401 a, 401 b, . . . 401 i, respectively. Additionally, the control circuit 403 generates a second group of enable signals EN2 a, EN2 b, . . . EN2 j for enabling the attenuator slices 402 a, 402 b, . . . 402 j, respectively. The number of SST slices i and the number of attenuator slices j can be the same or different. The control circuit 403 maintains a total number of active SST driver slices and active attenuator slices constant. Thus, for every differential SST driver slice that is disabled for amplitude control, an attenuator slice is enabled.
In certain implementations, each attenuator slice is implemented to have an on-state resistance about equal to an on-state resistance of one of the differential SST driver slices. For example, when operating at room temperature and nominal operating voltage, the on-state resistances can be within 20% of one another, or more particularly within 5%, for example, within 1%. Thus, the resistances of the attenuator slices and differential SST driver slices need not match exactly.
Although the attenuator slices and SST resistances can be implemented to be about equal in resistance, other implementations are possible. For example, making the attenuator resistance greater can provide increased amplitude control granularity.
In certain implementations, resistance tuning of the SST slices and/or attenuator slices can be provided to provide compensation for variation, for example, process, supply voltage, and/or temperature (PVT) variation.
Additionally or alternatively, the layouts and/or design topologies of the slices can be implemented such that the resistances of the SST slices and attenuator slices track each other to account for variation arising from manufacturing and/or operating conditions.
When the on-state resistances are about equal and the combined total number of active SST driver slices and active attenuator slices remains constant, the output impedance across attenuation settings remains constant at a desired value (for instance, 50 Ohms).
In certain implementations, the control circuit 403 can be implemented to also disable any clock and data path circuits used to drive a differential SST driver slice that is disabled for amplitude control. For example, in the illustrated embodiment, the data/clock path slices 404 a, 404 b, . . . 404 i also receive the enable signals EN1 a, EN1 b, . . . EN1 i.
FIG. 10A is a schematic diagram of another embodiment of a driver 500 with controllable swing and constant output impedance.
The driver 500 is depicted as including [0 . . . n-m-1] number of active SST slices 501 and corresponding data signals D and inverted data signals DB. Each of the SST slices 501 includes a first driver PFET 511, a first driver NFET 512, a second driver PFET 513, a second driver NFET 514, a first output resistor 515, and a second output resistor 516. The SST slices 501 have an output resistance Rsst. The SST slices 501 drive a pair of output terminals (to generate a differential output voltage +Vout/−Vout) between which a first 50 Ohm load resistor 505 and a second 50 Ohm load resistor 506 are connected.
With continuing reference to FIG. 10A, the driver 500 is depicted with [0 . . . m-1] number of active attenuator slices 502 connected differentially across the pair of output terminals. Each of the attenuator slices 502 includes a first resistor 521, a second resistor 522, and a T-gate multiplexer includes an NFET 523 and a PFET 524. The attenuator slices have an output resistance Rsst′.
In this embodiment, n represents the number of SST slices desired for nominal amplitude at 50 Ohm, and m is the number of decommissioned SST slices/activated attenuator slices.
FIG. 10B is a circuit diagram of a portion of the driver 500 of FIG. 10A. The circuit diagram depicts a pair of resistors of resistance R1 representing a transistor resistance of the SST slices. Additionally, the circuit diagram depicts a capacitor of capacitance Cx (which can included for any of the differential drivers herein for common mode termination) and a pair of resistors of resistance R2 representing the resistance of the attenuator slices. The diagram is annotated for a supply voltage (V_DD) of 1 V and a ground voltage of 0 V.
In this circuit, the differential output voltage (V_OUT+−V_OUT−) of the driver is equal to V_DD*(R2∥50 Ohm)/((R2∥50 Ohm)+R1).
FIG. 11 is a graph of one example of driver amplitude reduction versus attenuation setting for the driver 500 of FIGS. 10A and 10B. The x-axis represents the number of decommissioned SST slices and corresponding amplitude reduction. The output impedance is maintained at 50 Ohm across amplitude control settings.
FIG. 12A is a graph of one example comparison of driver output swing versus attenuation setting for two implementations of drivers. The graph includes a plot 601 for one implementation of the driver 500 of FIGS. 10A and 10B, and a second plot 602 for an array of selectable SST slices without any attenuator slices.
FIG. 12B is a graph of one example comparison of driver current versus attenuation setting for two implementations of drivers. The graph includes a plot 603 for one implementation of the driver 500 of FIGS. 10A and 10B, and a second plot 604 for an array of selectable SST slices without any attenuator slices.
FIG. 12C is a graph of one example comparison of driver output impedance versus attenuation setting for two implementations of drivers. The graph includes a plot 605 for one implementation of the driver 500 of FIGS. 10A and 10B, and a second plot 606 for an array of selectable SST slices without any attenuator slices.
With reference to FIGS. 12A to 12C, the total of SST slices and attenuator slices is selected to be 24 for the implementation of the driver 500. Additionally, a 100 Ohm differential resistance is simulated. The results show a linear tradeoff between 1V AVDD current reduction and smaller swing.
Moreover, with respect to FIG. 12A, the attenuation plot 601 follows FIG. 11 as expected while the attenuation plot 602 follows a more linear trend due to a different attenuation method. Additionally, the current plot 604 of FIG. 12B sees current increase as front-end stages ( slices 404 a, 404 b, . . . 404 i and slices 401 a, 401 b, . . . 401 i in FIG. 9 ) are still active while the attenuation plot 602 of FIG. 12A also induces additional cross-bar current within the output driver stage. Furthermore, with respect to FIG. 12C, the lower value of output impedance plot 606 is due to the 24 slices used in the simulation instead of 23 which would have increased the value by 4.3% (equal to 1/23) to 97.25 Ohm to match the output impedance plot 605 closer.
FIG. 13 is a schematic diagram of one embodiment of a SerDes system 700. The SerDes system 700 includes a first semiconductor die 701 and a second semiconductor die 702 connected over a high-speed link 703, which can be, for example, a pair of differential conductors.
The first semiconductor die 701 includes a serializer 704 that receives two or more incoming data streams of reduced bit rate relative to a high-speed data stream provided on the high-speed link 703. The second semiconductor die 702 includes a deserializer 705 that generates two or more outgoing data streams of reduced bit rate based on the high-speed data stream received from the serializer 704.
The serializer 704 includes a driver 706 implemented in accordance with one or more features of the present disclosure. For example, the driver 706 can include predriver circuitry 707 that provides multiplexing of the incoming data streams and/or can include SST swing control 708 using attenuator slices for output impedance control in accordance with the teachings herein.

Conclusion

The foregoing description may refer to elements or features as being “connected” or “coupled” together. As used herein, unless expressly stated otherwise, “connected” means that one element/feature is directly or indirectly connected to another element/feature, and not necessarily mechanically. Likewise, unless expressly stated otherwise, “coupled” means that one element/feature is directly or indirectly coupled to another element/feature, and not necessarily mechanically. Thus, although the various schematics shown in the figures depict example arrangements of elements and components, additional intervening elements, devices, features, or components may be present in an actual embodiment (assuming that the functionality of the depicted circuits is not adversely affected).
While certain embodiments have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the disclosure. Indeed, the novel apparatus, methods, and systems described herein may be embodied in a variety of other forms; furthermore, various omissions, substitutions and changes in the form of the methods and systems described herein may be made without departing from the spirit of the disclosure. For example, while the disclosed embodiments are presented in a given arrangement, alternative embodiments may perform similar functionalities with different components and/or circuit topologies, and some elements may be deleted, moved, added, subdivided, combined, and/or modified. Each of these elements may be implemented in a variety of different ways. Any suitable combination of the elements and acts of the various embodiments described above can be combined to provide further embodiments. Accordingly, the scope of the present invention is defined only by reference to the appended claims.
Although the claims presented here are in single dependency format for filing at the USPTO, it is to be understood that any claim may depend on any preceding claim of the same type except when that is clearly not technically feasible.

Claims

1-20. (canceled)

21. A driver circuit comprising:

a pair of output terminals configured to provide a differential output signal, the pair of output terminals comprising a first output terminal and a second output terminal;

a plurality of differential series source transistor (SST) driver slices electrically connected in parallel with one another and configured to drive the pair of output terminals, wherein each differential SST driver slice comprises a first driver p-type field effect transistor (PFET) connected between a high supply voltage and the first output terminal, a first driver n-type field effect transistor (NFET) connected between a low supply voltage and the first output terminal, a second driver PFET connected between the high supply voltage and the second output terminal, and a second driver NFET connected between the low supply voltage and the second output terminal;

a plurality of attenuator slices connected in parallel with one another across the pair of output terminals; and

a control circuit configured to selectively deactivate one or more of the differential SST driver slices to control an amplitude of the differential output signal, and to enable a corresponding number of the attenuator slices to provide output impedance compensation.

22. The driver circuit of claim 21, wherein each of the attenuator slices comprises a first resistor, a switch, and a second resistor in series.

23. The driver circuit of claim 22, further comprising a common-mode capacitor having one end connected between the first resistor and the second resistor.

24. The driver circuit of claim 22, wherein the switch comprises a p-type field effect transistor (PFET) and an n-type field effect transistor (NFET) in parallel.

25. The driver circuit of claim 21, wherein each attenuator slice has an on-state resistance within 20 percent of an on-state resistance of one of the differential SST driver slices.

26. The driver circuit of claim 21, wherein the driver circuit is operable in a plurality of attenuation settings each associated with deactivating a different number of the differential SST driver slices.

27. The driver circuit of claim 26, wherein the control circuit is configured to maintain a combined total number of active differential SST driver slices and active attenuator slices constant for each of the plurality of attenuation settings.

28. The driver circuit of claim 26, wherein an output impedance of the pair of output terminals is substantially constant across the plurality of attenuation settings.

29. (canceled)

30. The driver circuit of claim 21, further comprising a plurality of data and clock path slices coupled to the plurality of differential SST driver slices, wherein the control circuit is further configured to disable one or more of the plurality of data and clock path slices when deactivating the one or more of the differential SST driver slices.

31. A method of output swing control in a driver circuit, the method comprising:

providing a differential output signal on a pair of output terminals, the pair of output terminals comprising a first output terminal and a second output terminal;

driving the pair of output terminals using a plurality of differential series source transistor (SST) driver slices electrically connected in parallel with one another, wherein each differential SST driver slice comprises a first driver p-type field effect transistor (PFET) connected between a high supply voltage and the first output terminal, a first driver n-type field effect transistor (NFET) connected between a low supply voltage and the first output terminal, a second driver PFET connected between the high supply voltage and the second output terminal, and a second driver NFET connected between the low supply voltage and the second output terminal;

deactivating one or more of the differential SST driver slices to control an amplitude of the differential output signal; and

enabling a corresponding number of a plurality of attenuator slices to provide output impedance compensation, wherein the plurality of attenuator slices are connected in parallel with one another across the pair of output terminals.

32. The method of claim 31, further comprising providing common-mode termination to the pair of output terminals using a first resistor and a second resistor connected in series between the pair of output terminals and a common-mode capacitor having one end connected between the first resistor and the second resistor.

33. The method of claim 31, wherein each attenuator slice has an on-state resistance within 20 percent of an on-state resistance of one of the differential SST driver slices.

34. The method of claim 31, further comprising maintaining a combined total number of active differential SST driver slices and active attenuator slices constant for each of a plurality of attenuation settings of the differential output signal.

35. The method of claim 31, further comprising maintaining an output impedance of the pair of output terminals substantially constant across the plurality of attenuation settings.

36. The method of claim 31, further comprising providing data to the plurality of differential SST driver slices using a plurality of data and clock path slices, and disabling one or more of the plurality of data and clock path slices when deactivating the one or more of the differential SST driver slices.

37. A serializer/deserializer (SerDes) system comprising:

a deserializer; and

a serializer comprising a driver including a pair of output terminals configured to provide a differential output signal to the deserializer, a plurality of differential series source transistor (SST) driver slices electrically connected in parallel with one another and configured to drive the pair of output terminals, a plurality of attenuator slices connected in parallel with one another across the pair of output terminals, and a control circuit configured to selectively deactivate one or more of the differential SST driver slices to control an amplitude of the differential output signal, and to enable a corresponding number of the attenuator slices to provide output impedance compensation, wherein the pair of output terminals comprises a first output terminal and a second output terminal, wherein each differential SST driver slice comprises a first driver p-type field effect transistor (PFET) connected between a high supply voltage and the first output terminal, a first driver n-type field effect transistor (NFET) connected between a low supply voltage and the first output terminal, a second driver PFET connected between the high supply voltage and the second output terminal, and a second driver NFET connected between the low supply voltage and the second output terminal.

38. The SerDes system of claim 37, wherein the driver circuit is operable in a plurality of attenuation settings each associated with deactivating a different number of the differential SST driver slices.

39. The SerDes system of claim 38, wherein the control circuit is configured to maintain a combined total number of active SST driver slices and active attenuator slices constant for each of the plurality of attenuation settings.

40. The SerDes system of claim 38, wherein an output impedance of the pair of output terminals is substantially constant across the plurality of attenuation settings.

41. The SerDes system of claim 38, wherein each of the attenuator slices comprises a first resistor, a switch, and a second resistor in series, and a common-mode capacitor having one end connected between the first resistor and the second resistor.