Ultra high-speed transmitter and storage device including the same
The ultra-high-speed transmitter design addresses reduced timing margins in PAM-2 signaling by using a feedback circuit to adjust setup and hold margins with multi-phase clock signals, ensuring stable data transmission despite process, voltage, and temperature variations.
Patent Information
- Application Number
- KR1020220156033
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-11-21
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2042-11-21
AI Technical Summary
Conventional PAM-2 signaling in semiconductor integrated circuits faces challenges with reduced setup/hold timing margins due to increased data transmission rates, process, voltage, and temperature variations, leading to channel attenuation and clock quality degradation.
An ultra-high-speed transmitter design incorporating a pattern generator, data generator, serializer, transmission driver, and feedback circuit to adjust setup and hold margins using multi-phase clock signals, ensuring stable data transmission by detecting and controlling these margins in real-time.
The design stabilizes data transmission by maintaining target setup and hold margins, even with process, voltage, and temperature changes, preventing degradation of jitter characteristics and ensuring reliable data transfer.
Smart Images

Figure R1020220156033_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to a semiconductor integrated circuit, and more specifically, to an ultra-high-speed transmitter capable of securing a timing margin and a storage device including the same. Background Technology
[0002] Recently, there has been a growing demand for memory bandwidth for High Performance Computing (HPC), such as Artificial Intelligence (AI) and Graphic Processing Units (GPUs). Expanding bandwidth can depend on innovations in process technology, and advancements in process technology have created higher densities in Integrated Circuits (ICs).
[0003] To expand the bandwidth mentioned above, 2-level pulse amplitude modulation (PULSE AMPLITUDE MODULATION-2: PAM-2) signaling, which generally has two data modulation levels, is used. Conventional PAM-2 signaling has the advantage of simplifying the structure of the transmitter and receiver. However, only one bit can be transmitted during one unit interval (UI), and the clock frequency must be increased to increase the bandwidth, which leads to problems such as channel attenuation and clock quality degradation.
[0004] In addition, for the transmitter to operate normally, there must be a sufficient setup / hold timing margin between the clock signal and the data signal. However, as the transmitter's data transmission rate increases, the setup / hold timing margin decreases. The problem to be solved
[0005] One objective of the present invention is to provide an ultra-high-speed transmitter capable of providing sufficient setup / hold timing margins even with process, voltage, and temperature changes.
[0006] One objective of the present invention is to provide a storage device including an ultra-high-speed transmitter capable of providing sufficient setup / hold timing margin even with process, voltage, and temperature changes. means of solving the problem
[0007] To achieve the above objective, an ultra-high-speed transmitter according to embodiments of the present invention comprises a pattern generator, a data generator, a serializer, a transmission driver, and a feedback circuit. The pattern generator generates test data having a specific pattern. The data generator generates a retimed data signal and retimed test data having a controlled setup margin and a hold margin by controlling the delay amount of each of the input data signal and the test data based on controlled clock signals. The serializer serializes the retimed data signal based on multi-phase clock signals to generate a serialized data signal. The transmission driver generates an output data signal based on the serialized data and transmits the output data signal through a channel. The feedback circuit detects the setup margin and the hold margin of the retimed test data through a path separate from the path of the retimed data signal, and generates the controlled clock signals by controlling the delay amount of the multi-phase clock signals based on the detected setup margin and the hold margin.
[0008] To achieve the above objective, a storage device according to embodiments of the present invention includes a processor and an ultra-high-speed transmitter. The processor generates an input data signal. The ultra-high-speed transmitter generates test data and, based on the test data, adjusts the delay amount of the input data signal to transmit an output data signal having a target setup margin and a target hold margin to a host through a link. The ultra-high-speed transmitter includes a pattern generator, a data generator, a serializer, a transmission driver, and a feedback circuit. The pattern generator generates the test data having a specific pattern. The data generator generates a retimed data signal and retimed test data having adjusted setup margins and hold margins by adjusting the delay amounts of the input data signal and the test data, respectively, based on adjusted clock signals. The serializer serializes the retimed data signal based on multi-phase clock signals to generate a serialized data signal. The transmission driver generates an output data signal based on the serialized data and transmits the output data signal through a channel. The feedback circuit detects the setup margin and the hold margin of the retimed test data through a path separate from the path of the retimed data signal, and generates the adjusted clock signals by adjusting the delay amount of the multiphase clock signals based on the detected setup margin and the hold margin.
[0009] To achieve the above objective, an ultra-high-speed transmitter according to embodiments of the present invention comprises a pattern generator, a data generator, a serializer, a transmission driver, and a feedback circuit. The pattern generator generates test data having a specific pattern. The data generator generates a retimed data signal and retimed test data having a controlled setup margin and a hold margin by controlling the delay amount of each of the input data signal and the test data based on controlled clock signals. The serializer serializes the retimed data signal based on multi-phase clock signals to generate a serialized data signal. The transmission driver generates an output data signal based on the serialized data and transmits the output data signal through a channel. The feedback circuit detects the setup margin and the hold margin of the retimed test data through a path separate from the path of the retimed data signal, and generates the controlled clock signals by controlling the delay amount of the multi-phase clock signals based on the detected setup margin and the hold margin. The feedback circuit includes a first dummy multiplexer, a second dummy multiplexer, a sensing circuit, control logic, and a clock timing regulator. The first dummy multiplexer sequentially selects a first portion of the test bits of the retimed test data based on the multiphase clock signals and provides it as a first output signal. The second dummy multiplexer sequentially selects a second portion of the test bits of the retimed test data, which is different from the first portion, based on the multiphase clock signals and provides it as a second output signal. The sensing circuit detects the setup margin and the hold margin of the retimed test data based on the first output signal, the second output signal, and a strobe signal selected among the test bits, and generates a sensing signal indicating the detected setup margin and the hold margin.The above control logic generates a delay control signal based on the detection signal. The clock timing regulator generates the adjusted clock signals by adjusting the delay amount of the multi-phase clock signals based on the delay control signal. Effects of the invention
[0010] An ultra-high-speed transmitter and a storage device including the same according to embodiments of the present invention can stably transmit a data signal even when the data transmission rate increases by adjusting the setup margin and hold margin of the test data in real time based on test data and multi-phase clock signals transmitted through a path separate from the path through which the data signal is transmitted, determining a lock point having a target setup margin and a target hold margin based on the adjusted setup margin and hold margin, and transmitting a data signal having a delay at the lock point. Brief explanation of the drawing
[0011] FIG. 1 is a block diagram showing a transmitter and a receiver according to embodiments of the present invention. FIG. 2 is a block diagram showing an apparatus including a transmitter according to embodiments of the present invention. FIG. 3 is a block diagram showing an example of the transmitter of FIG. 1 according to embodiments of the present invention. FIG. 4 is a block diagram showing an example of a data generator in the transmitter of FIG. 3 according to embodiments of the present invention. FIG. 5 is a block diagram showing an example of a serializer in the transmitter of FIG. 3 according to embodiments of the present invention. FIG. 6 is a circuit diagram showing an example of a transmission driver in the transmitter of FIG. 3 according to embodiments of the present invention. FIG. 7 is a block diagram showing an example of a clock timing regulator in the transmitter of FIG. 3 according to embodiments of the present invention. FIG. 8 shows a first dummy multiplexer and a second dummy multiplexer in the transmitter of FIG. 3 according to embodiments of the present invention. FIG. 9 is a circuit diagram showing an example of a sensing circuit in the transmitter of FIG. 3 according to embodiments of the present invention. FIG. 10a shows test bits of test data provided by the pattern generator of the transmitter of FIG. 3, and FIG. 10b shows multiphase provided to the clock timing regulator Figure 10c represents the output data signal in Figure 3. Figures 11 through 13 represent the skew between the multiphase clock signals and test bits in the dummy multiplexers of Figure 8 and the sensing circuit of Figure 9. Figure 14a represents the first voltage signal, the second voltage signal, and the sensing signal in the sensing circuit of Figure 9 according to embodiments of the present invention. Figure 14b represents the first voltage signal and the second voltage signal near the lock point of Figure 14a. Figure 15 represents the skew between the multiphase clock signals and test bits at the lock point. Figures 16a through 16d each represent that the lock point changes according to the strobe signal selected by the strobe signal selector in the feedback circuit of Figure 3. Figure 17 is a block diagram representing a memory system according to embodiments of the present invention. Figures 18 and 19 are block diagrams representing examples of a system according to embodiments of the present invention.FIG. 20 is a block diagram illustrating an example of a semiconductor memory device included in the system of FIG. 18 and FIG. 19 according to embodiments of the present invention. FIG. 21 illustrates a first bank array in the semiconductor memory device of FIG. 20 according to embodiments of the present invention. FIG. 22 is a block diagram illustrating a storage system according to embodiments of the present invention. FIG. 23 and FIG. 24 are drawings illustrating an example of an interconnect architecture applied to a storage system according to embodiments of the present invention. FIG. 25 is a block diagram illustrating a storage system according to embodiments of the present invention. FIG. 26 is an exemplary block diagram showing a semiconductor memory device according to embodiments of the present invention. FIG. 27 is a structural diagram illustrating an example of a semiconductor package including a stacked memory device according to embodiments of the present invention. Specific details for implementing the invention
[0012] Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the attached drawings. Identical components in the drawings are given the same reference numerals, and redundant descriptions of identical components are omitted.
[0013] Multi-level signaling can be used as a means to compress the bandwidth required to transmit data at a given bit rate. In a simple binary system, two voltage levels are typically used to represent 1 and 0, and the symbol rate can be equal to the bit rate. In contrast, multi-level signaling uses k (where k is a natural number greater than or equal to 3) symbols to represent data, allowing each symbol to represent more than 1 bit of data. Consequently, the symbol rate is smaller than the bit rate, and thus the bandwidth can be compressed.
[0014] In other words, multi-level signaling can be used to increase the data transmission rate without increasing the data transmission frequency or power. An example of such multi-level signaling is pulse amplitude modulation (hereinafter 'PAM'), in which a multi-level signal can represent multiple bits of data.
[0015] FIG. 1 is a block diagram showing a transmitter and a receiver according to embodiments of the present invention.
[0016] Referring to FIG. 1, the transmitter (100) and the receiver (200) can be connected to each other through a channel or link (30).
[0017] The transmitter (100) can generate an output data signal (TX_OUT) having a plurality of PAM-3 symbols, each having a first voltage level (VL1), a second voltage level (VL2), and a third voltage level (VL3), based on an input data signal (TX_IN) having a plurality of binary input bits and multiphase clock signals (MPCK). The output data signal (TX_OUT) is transmitted to a receiver (20) through a channel (30). The receiver (200) can generate an output data signal (RX_OUT) having a plurality of binary output bits based on an input data signal (RX_IN) having a plurality of PAM-3 symbols and multiphase clock signals (MPCK).
[0018] The three voltage levels of the output data signal (TX_OUT) may include a first voltage level (VL1), a second voltage level (VL2) higher than the first voltage level (VL1), and a third voltage level (VL3) higher than the second voltage level (VL2). The first voltage level (VL1), the second voltage level (VL2), and the third voltage level (VL3) may be referred to as the low level, the middle (or mid) level, and the high level, respectively.
[0019] The output data signal (RX_OUT) may have two different voltage levels, and one value (or bit) included in the output data signal (RX_OUT) may represent one value (or bit) included in the input data signal (TX_IN).
[0020] In the embodiment, the output data signal (TX_OUT) and the input data signal (RX_IN) may each have two different voltage levels.
[0021] FIG. 2 is a block diagram showing a device including a transmitter according to embodiments of the present invention.
[0022] Referring to FIG. 2, the device (70) may include a processor (80), a clock generator (90), and a transmitter (100).
[0023] The processor (80) can generate an input data signal (TX_IN) and provide the input data signal (TX_IN) to the transmitter (100). The clock generator (90) can generate multiphase clock signals (MPCK) and provide the multiphase clock signals (MPCK) to the transmitter (100). In an embodiment, the clock generator (90) can be implemented as a phase-locked loop (PLL) circuit, etc.
[0024] The transmitter (100) can convert an input data signal (TX_IN) into an output data signal (TX_OUT) based on multi-phase clock signals (MPCK) and transmit the output data signal (TX_OUT) through a channel or link.
[0025] The input data signal (TX_IN) is a parallel data signal, and the output data signal (TX_OUT) can be a serial data signal.
[0026] In the embodiment, the device (70) may be a volatile memory device or a storage device including a non-volatile memory device.
[0027] FIG. 3 is a block diagram showing an example of the transmitter of FIG. 1 according to embodiments of the present invention.
[0028] Referring to FIG. 3, the transmitter (100) may include a pattern generator (110), a data generator (120), a serializer (150), a transmission driver (170), and a feedback circuit (200).
[0029] The pattern generator (110) can generate test data (TD) having a certain pattern.
[0030] The data generator (120) receives an input data signal (RX_IN) and test data (TD), and can generate a retimed data signal (TX_RT) and a retimed test data (TD_RT) having a controlled setup margin and a hold margin, respectively, by controlling the delay amount of each of the input data signal (RX_IN) and test data (TD) based on controlled clock signals (ACK).
[0031] The serializer (150) can serialize a retimed data signal (TX_RT) based on multiphase clock signals (MPCK) to generate a serialized data signal (TX_SER) and provide the serialized data signal (TX_SER) to a transmission driver (170).
[0032] The transmitting driver (170) can generate an output data signal (TX_OUT) based on a serialized data signal (TX_SER) and transmit the output data signal (TX_OUT) through a link or channel.
[0033] The feedback circuit (200) can detect the setup margin and hold margin of the retimed test data (TD_RT) through a path separate from the path of the retimed data signal (TX_RT), and based on the detected setup margin and hold margin, adjust the delay amount of the multiphase clock signals (MPCK) to generate the adjusted clock signals (ACK), and provide the adjusted clock signals (ACK) to the data generator (120).
[0034] The feedback circuit (200) can adjust the delay amount of the multiphase clock signals (MPCK) until the retimed test data (TD_RT) has a target setup margin and a target hold margin, and when the retimed test data (TD_RT) has a target setup margin and a target hold margin, it can fix (lock) the delay amount of the multiphase clock signals (MPCK).
[0035] The data generator (120) generates a retimed data signal (TX_RT) from an input data signal (RX_IN) based on multiphase clock signals (MPCK) having a fixed delay amount, so the retimed data signal (TX_RT) can have a target setup margin and a target hold margin, and even if the data transmission rate of the output data signal (TX_OUT) increases, the output data signal (TX_OUT) can have jitter characteristics that are not degraded with respect to at least one of changes in process, voltage and temperature.
[0036] The feedback circuit (200) may include a clock timing regulator (210), a first dummy multiplexer (220), a second dummy multiplexer (230), a sensing circuit (240), control logic (270), and a strobe signal selector (280).
[0037] The first dummy multiplexer (220) can sequentially select a first portion of the test bits of the retimed test data (TD_RT) based on multiphase clock signals (MPCK) and provide it as a first output signal (OUT1). The second dummy multiplexer (230) can sequentially select a second portion of the test bits of the retimed test data (TD_RT) different from the first portion based on multiphase clock signals (MPCK) and provide it as a second output signal (OUT2).
[0038] The detection circuit (240) can detect the setup margin and hold margin of the retimed test data (TD_RT) based on the first output signal (OUT1), the second output signal (OUT2), and the strobe signal (STR), generate a detection signal (DS) representing the detected setup margin and hold margin, and provide the detection signal (DS) to the control logic (270).
[0039] The control logic (270) can generate a delay control signal (DCTL) based on the detection signal (DS) and provide the delay control signal (DCTL) to the clock timing regulator (210). The control logic (270) can also provide a select signal (SEL) to the strobe signal selector (280).
[0040] The clock timing regulator (210) receives multiphase clock signals (MPCK) and a delay control signal (DCTL), adjusts the delay amount of the multiphase clock signals (MPCK) to generate adjusted clock signals (ACK), and can provide the adjusted clock signals (ACK) to the data generator (120).
[0041] The strobe signal selector (280) receives retimed test data (TD_RT), selects one of the test bits of the retimed test data (TD_RT) based on a selection signal (SEL), and can provide the selected one to the detection circuit (240) as a strobe signal (STR).
[0042] FIG. 4 is a block diagram showing an example of a data generator in the transmitter of FIG. 3 according to embodiments of the present invention.
[0043] Referring to FIG. 4, the data generator (120) may include a first shift register block (130) and a second shift register block (140).
[0044] The first shift register block (130) may include a plurality of first shift registers (130a, 130b, 130c, 130d). The plurality of first shift registers (130a, 130b, 130c, 130d) may delay the data bits of an input data signal (TX_IN) in parallel based on a first adjusted clock signal (ACK1) and a second adjusted clock signal (ACK3) having a phase difference of 180 degrees among the adjusted clock signals (ACK), thereby generating a retimed data signal (TX_RT).
[0045] Among the plurality of first shift registers (130a, 130b, 130c, 130d), the shift register (130a) may include a plurality of first latches (131, 132, 133, …, 13k, where k is an even natural number greater than 3). The plurality of first latches (131, 132, 133, …, 13k) are cascaded to sequentially latch corresponding data bits among the data bits of the input data signal (TX_IN). Among the plurality of first latches (131, 132, 133, …, 13k), the first odd latches (131, 133) latch a data bit input based on the first regulated clock signal (ACK1), and among the plurality of first latches (131, 132, 133, …, 13k), the first even latches (132, 13k) latch a data bit input based on the second regulated clock signal (ACK3).
[0046] The configuration of each of the shift registers (130b, 130c, 130d) may be the same as the configuration of the shift register (130a).
[0047] Among the plurality of second shift registers (140a, 140b, 140c, 140d), the shift register (140a) may include a plurality of second latches (141, 142, 143, …, 14k). The plurality of second latches (141, 142, 143, …, 14k) are cascaded to sequentially latch corresponding test bits among the test bits of test data (TD_IN). Among the plurality of second latches (141, 142, 143, …, 14k), the second odd latches (141, 143) latch a test bit input based on the first regulated clock signal (ACK1), and among the plurality of second latches (141, 142, 143, …, 14k), the second even latches (142, 14k) latch a test bit input based on the second regulated clock signal (ACK3).
[0048] The configuration of each of the shift registers (140b, 140c, 140d) may be the same as the configuration of the shift register (140a).
[0049] FIG. 5 is a block diagram showing an example of a serializer in the transmitter of FIG. 3 according to embodiments of the present invention.
[0050] Referring to FIG. 5, the serializer (150) may include a plurality of multiplexers (150a, 150b, 150c, 150d).
[0051] Multiple multiplexers (150a, 150b, 150c, 150d) operate in parallel and can sequentially select data bits of a retimed data signal (TX_RT) based on multiphase clock signals (MPCK) to generate a serialized data signal (TX_SER).
[0052] FIG. 6 is a circuit diagram showing an example of a transmitting driver in the transmitter of FIG. 3 according to embodiments of the present invention.
[0053] Referring to FIG. 6, the transmitting driver (170) may include a plurality of driver segments (170a, 170b, 170c, 170d).
[0054] Multiple driver segments (170a, 170b, 170c, 170d) commonly receive a first sub-data signal (TX_SER_P) and a second sub-data signal (TX_SER_N) of a serialized data signal (TX_SER) and can be connected in parallel between a power supply voltage (VDD) and a ground voltage (VSS).
[0055] FIG. 6 shows the configuration of driver segment (170a) among multiple driver segments (170a, 170b, 170c, 170d), but the configurations of 170b, 170c, and 170d may be substantially identical to the configuration of driver segment (170a). FIG. 6 also assumes that the serialized data signal (TX_SER) is a differential signal.
[0056] The driver segment (170a) may include a PMOS transistor (171) and first to fifth N-MOS transistors (172, 173, 174, 175).
[0057] The PMOS transistor (171) is connected between the power supply voltage (VDD) and the first node (N11) and may have a gate that receives a corresponding bit (PUCD1) of the pull-up control code (PUCD).
[0058] The first N-MOS transistor (172) may be connected between the first node (N11) and the second node (N12) and may have a gate that receives the first sub-data signal (TX_SER_P). The second N-MOS transistor (173) may be connected in parallel with the first N-MOS transistor (172) between the first node (N11) and the third node (N13) and may have a gate that receives the second sub-data signal (TX_SER_N). The third N-MOS transistor (174) may be connected between the second node (N12) and the fourth node (N14) and may have a gate that receives the second sub-data signal (TX_SER_N). The fourth N-MOS transistor (175) may be connected in parallel with the third N-MOS transistor (473) between the third node (N13) and the fourth node (N14) and may have a gate that receives the first sub-data signal (TX_SER_P).
[0059] The fifth N-MOS transistor (176) may be connected between the fourth node (N14) and the ground voltage (VSS) and may have a gate that receives a corresponding bit (PDCD1) of a pull-down control code (PDCD).
[0060] At the second node (N12), the first component (TX_OUT_P) of the output data signal (TX_OUT) is provided to the first transmission pad (TXPD) and transmitted to an external device through the link (30), and at the third node (N13), the second component (TX_OUT_N) of the output data signal (TX_OUT) is provided to the second transmission pad (TXND) and transmitted to an external device through the link (30).
[0061] When driver segment (170a) among driver segments (170a, 170b, 170c, 170d) is activated by a pull-up control code (PUCD) and a pull-down control code (PDCD), the transmitting driver (170) can provide an impedance having a first impedance value by connecting to a PMOS transistor (171) and first to fifth NMOS transistors (172, 173, 174, 175, 176).
[0062] When driver segments (170a, 170b) among driver segments (170a, 170b, 170c, 170d) are activated by a pull-up control code (PUCD) and a pull-down control code (PDCD), the transmitting driver (170) can provide an impedance having a second impedance value, which is formed by connecting two impedances having a first impedance value in parallel to the PMOS transistor (171) and the first to fifth NMOS transistors (172, 173, 174, 175, 176) of the driver segments (170a, 170b).
[0063] The pull-up control code (PUCD) and pull-down control code (PDCD) can be provided from the processor (80) of FIG. 2.
[0064] FIG. 7 is a block diagram showing an example of a clock timing regulator in the transmitter of FIG. 3 according to embodiments of the present invention.
[0065] Referring to FIG. 7, the clock timing regulator (210) may include a plurality of voltage-controlled delay lines (VCDL) (211, 212, 213, 214).
[0066] Each of the voltage-controlled delay lines (211, 212, 213, 214) can provide adjusted clock signals (ACK1, ACK2, ACK3, ACK4) by adjusting the delay amount of each of the clock signals (CK1, CK2, CK3, CK4) of the multiphase clock signals (MPCK) in response to the delay control signal (DCTL).
[0067] In the embodiment, each of the voltage control delay lines (211, 212, 213, 214) may be composed of a plurality of cascade-connected inverters.
[0068] In an embodiment, the clock timing regulator (210) may include a plurality of digitally-controlled delay lines (DCDL).
[0069] FIG. 8 shows a first dummy multiplexer and a second dummy multiplexer in the transmitter of FIG. 3 according to embodiments of the present invention.
[0070] FIG. 8 shows that the test data (TD) of FIG. 3 includes first to fourth test bits (Td1, Td2, Td3, Td4), and that the first dummy multiplexer (220) and the second dummy multiplexer (230) operate based on the first clock signal (CK1) and the second clock signal (CK2), which have a phase difference of 90 sets among the multiphase clock signals (MPCK).
[0071] Referring to FIG. 8, the first dummy multiplexer (220) can sequentially select a first part of the first test bit (Td1), the second test bit (Td2), the third test bit (Td3), and the fourth test bit (Td4) according to the logic levels of the first clock signal (CK1) and the second clock signal (CK2) to provide a first output signal (OUT1), and the second dummy multiplexer (230) can sequentially select a second part different from the first part of the second test bit (Td2), the third test bit (Td3), the fourth test bit (Td4), and the first test bit (Td1) according to the logic levels of the first clock signal (CK1) and the second clock signal (CK2) to provide a second output signal (OUT2).
[0072] In an embodiment, the first dummy multiplexer (220) sequentially selects a first portion of test data (TD) according to the logic levels of the first to fourth clock signals (CK1, CK2, CK3, CK4) to provide a first output signal (OUT1), and the first dummy multiplexer (220) may sequentially select a second portion different from the first portion of test data (TD) according to the logic levels of the first to fourth clock signals (CK1, CK2, CK3, CK4) to provide a second output signal (OUT2).
[0073] FIG. 9 is a circuit diagram showing an example of a sensing circuit in the transmitter of FIG. 3 according to embodiments of the present invention.
[0074] Referring to FIG. 9, the sensing circuit (240) may include a first exclusive OR (XOR) gate (241), a second exclusive OR gate (243), a first integrator (245), a second integrator (247), and a comparator (250).
[0075] The first exclusive OR gate (241) can provide a third output signal (OUT3) by performing an exclusive OR operation on the first output signal (OUT1) and the strobe signal (STR). The second exclusive OR gate (243) can provide a fourth output signal (OUT4) by performing an exclusive OR operation on the second output signal (OUT2) and the strobe signal (STR).
[0076] The first integrator (245) can integrate the third output signal (OUT3) to output a first voltage signal (VOUT1) which is an analog signal. The second integrator (247) can integrate the fourth output signal (OUT4) to output a second voltage signal (VOUT2) which is an analog signal.
[0077] Since the first integrator (245) integrates the third output signal (OUT3) to output a first voltage signal (VOUT1) which is an analog signal, the third output signal (OUT3) may have a first duty cycle representing a first skew between the retimed test data (TD_RT) and the multiphase clock signals (MPCK). Since the second integrator (247) integrates the fourth output signal (OUT4) to output a second voltage signal (VOUT2) which is an analog signal, the fourth output signal (OUT4) may have a second duty cycle representing a second skew between the retimed test data (TD_RT) and the multiphase clock signals (MPCK).
[0078] The comparator (250) can output a detection signal (DS) by subtracting the second voltage signal (VOUT2) from the first voltage signal (VOUT1). Accordingly, the detection signal (DS) can have a magnitude and slope based on the magnitude of the first voltage signal (VOUT1) and the magnitude of the second voltage signal (VOUT2).
[0079] The control logic (270) of FIG. 3 can determine at least one point among the points where the first voltage signal (VOUT1) and the second voltage signal (VOUT2) intersect based on the detection signal (DS) where the slope of the detection signal (DS) is positive as a lock point. The retimed data signal (TX_RT) at the lock point may have a target setup margin and a target hold margin.
[0080] The lock point may change as the strobe signal (STR) provided by the strobe signal selector (280) of FIG. 3, selected from the test bits of the retimed test data (TD_RT) based on the selection signal (SEL), changes.
[0081] In response to the detection signal (DS) indicating that the first voltage signal (VOUT1) is greater than the second voltage signal (VOUT2), the control logic (270) of FIG. 3 generates a delay control signal (DCTL) to reduce the delay amount of the regulated clock signals (ACK), and the clock timing regulator (210) of FIG. 3 can generate the regulated clock signals (ACK) by reducing the delay amount of the multiphase clock signals (MPCK) in response to the delay control signal (DCTL).
[0082] In response to the detection signal (DS) indicating that the first voltage signal (VOUT1) is smaller than the second voltage signal (VOUT2), the control logic (270) of FIG. 3 generates a delay control signal (DCTL) to increase the delay amount of the regulated clock signals (ACK), and the clock timing regulator (210) of FIG. 3 can generate the regulated clock signals (ACK) by increasing the delay amount of the multiphase clock signals (MPCK) in response to the delay control signal (DCTL).
[0083] The control logic (270) adjusts the delay control signal (DCTL) so that the delay amount is adjusted until the detection signal (DS) indicates that the first voltage signal (VOUT1) is the same as the second voltage signal (VOUT2), and can lock the delay control signal (DCTL) in response to the first voltage signal (VOUT1) being the same as the second voltage signal (VOUT2).
[0084] FIG. 10a shows the test bits of the test data provided by the pattern generator of the transmitter of FIG. 3, FIG. 10b shows the multiphase clock signals provided to the clock timing regulator, and FIG. 10c shows the output data signal in FIG. 3.
[0085] Referring to FIG. 3, FIG. 10a, FIG. 10b, and FIG. 10c, when the time interval representing one value in the output data signal (TX_OUT) as illustrated in FIG. 10c is 1 unit interval (UI) (1UI), the time interval representing one value in the first, second, third, and fourth test bits (Td1, Td2, Td3, Td4) as illustrated in FIG. 10a and 10b and the period of the first, second, third, and fourth clock signals (CK1, CK2, CK3, CK4) may correspond to 4 unit intervals (4UI). Additionally, the input data signal (TX_IN) may also have the same time interval as the test bits (Td1, Td2, Td3, Td4).
[0086] FIGS. 11 to 13 show the skew between multiphase clock signals and test bits in the dummy multiplexers of FIG. 8 and the sensing circuit of FIG. 9.
[0087] In FIGS. 11 to 13, it is assumed that among the multiphase clock signals (MPCK), a first clock signal (CK1) and a second clock signal (CK2) having a phase difference of 90 degrees from each other are applied to a first dummy multiplexer (220) and a second dummy multiplexer (230), and a second test bit (Td2) is selected as a strobe signal (STR).
[0088] Referring to FIG. 11, the rising edge of the first test bit (Td1) may have a skew (SK1) corresponding to 1 UI with the falling edge of the second clock signal (CK2), the second test bit (Td2) is delayed by 1 UI with respect to the first test bit (Td1), the third test bit (Td3) is delayed by 1 UI with respect to the second test bit (Td2), and the fourth test bit (Td4) is delayed by 1 UI with respect to the third test bit (Td3).
[0089] The first dummy multiplexer (220) can sequentially select one of the test bits (Td1, Td2, Td3, Td4) based on the logic levels of the first clock signal (CK1) and the second clock signal (CK2) and provide it as the first output signal (OUT1), and the second dummy multiplexer (230) can sequentially select the other of the test bits (Td2, Td3, Td4, Td1) based on the logic levels of the first clock signal (CK1) and the second clock signal (CK2) and provide it as the second output signal (OUT1).
[0090] The first exclusive OR gate (241) can perform an exclusive OR operation on the first output signal (OUT1) and the strobe signal (STR) to output a third output signal (OUT3), and the first integrator (245) can provide a first voltage signal (VOUT1) by integrating the third output signal (OUT3). The second exclusive OR gate (243) can perform an exclusive OR operation on the second output signal (OUT3) and the strobe signal (STR) to output a fourth output signal (OUT4), and the second integrator (247) can provide a second voltage signal (VOUT2) by integrating the fourth output signal (OUT4).
[0091] Accordingly, the first output signal (OUT1), the second output signal (OUT2), the third output signal (OUT3), the fourth output signal (OUT4), the first voltage signal (VOUT1) and the second voltage signal (OUT2) may have the illustrated waveforms.
[0092] The comparator (250) can output a detection signal (DS) by subtracting the second voltage signal (VOUT2) from the first voltage signal (VOU1).
[0093] Referring to FIG. 12, the rising edge of the first test bit (Td1) may have a skew (SK2) corresponding to 2UI with the falling edge of the second clock signal (CK2), the second test bit (Td2) is delayed by 1UI with respect to the first test bit (Td1), the third test bit (Td3) is delayed by 1UI with respect to the second test bit (Td2), and the fourth test bit (Td4) is delayed by 1UI with respect to the third test bit (Td3).
[0094] Accordingly, the first output signal (OUT1), the second output signal (OUT2), the third output signal (OUT3), the fourth output signal (OUT4), the first voltage signal (VOUT1) and the second voltage signal (OUT2) may have the illustrated waveforms.
[0095] The comparator (250) can output a detection signal (DS) by subtracting the second voltage signal (VOUT2) from the first voltage signal (VOU1).
[0096] Referring to FIG. 13, the rising edge of the first test bit (Td1) may have a skew (SK2) corresponding to 3.5 UI with respect to the falling edge of the second clock signal (CK2), the second test bit (Td2) is delayed by 1 UI with respect to the first test bit (Td1), the third test bit (Td3) is delayed by 1 UI with respect to the second test bit (Td2), and the fourth test bit (Td4) is delayed by 1 UI with respect to the third test bit (Td3).
[0097] Accordingly, the first output signal (OUT1), the second output signal (OUT2), the third output signal (OUT3), the fourth output signal (OUT4), the first voltage signal (VOUT1) and the second voltage signal (OUT2) may have the illustrated waveforms.
[0098] The comparator (250) can output a detection signal (DS) by subtracting the second voltage signal (VOUT2) from the first voltage signal (VOU1).
[0099] FIG. 14a shows a first voltage signal, a second voltage signal, and a detection signal in the detection circuit of FIG. 9 according to embodiments of the present invention.
[0100] Referring to FIG. 14a, it can be seen that among the points where the first voltage signal (VOUT1) and the second voltage signal (VOUT2) intersect, the points where the slope of the detection signal (DS) is positive are determined as lock points (LP1, LP2).
[0101] In FIG. 14a, the horizontal axis represents skew, and the vertical axis represents the voltage of the first voltage signal (VOUT1), the second voltage signal (VOUT2), and the detection signal (DS), respectively.
[0102] FIG. 14b shows a first voltage signal and a second voltage signal near the lock point of FIG. 14a.
[0103] Referring to FIG. 14b, it can be seen that after the lock point (LP1), the difference between the first voltage signal (VOU1) and the second voltage signal (VOUT2) is maintained within a constant range between the first level (L11) and the second level (L12).
[0104] Figure 15 shows the skew between the multiphase clock signals and test bits at the lock point.
[0105] Referring to FIG. 15, it can be seen that the rising edge of the first test bit (Td1) may have a skew corresponding to 1.5 UI with respect to the falling edge of the second clock signal (CK2), the second test bit (Td2) is delayed by 1 UI with respect to the first test bit (Td1), the third test bit (Td3) is delayed by 1 UI with respect to the second test bit (Td2), and the fourth test bit (Td4) is delayed by 1 UI with respect to the third test bit (Td3).
[0106] Accordingly, each of the test bits (Td1, Td2, Td3, Td4) of the retimed test data (TD_RT) can have a target setup margin (Tsu) corresponding to 1.5 UI and a target hold margin (Thd) corresponding to 1.5 UI. Accordingly, since the retimed data signal (TX_RT) has the same setup margin and hold margin as the retimed test data (TD_RT), the retimed data signal (TX_RT) can also have a target setup margin (Tsu) corresponding to 1.5 UI and a target hold margin (Thd) corresponding to 1.5 UI.
[0107] FIGS. 16a to 16d each show that the lock point changes depending on the strobe signal selected by the strobe signal selector in the feedback circuit of FIG. 3.
[0108] In each of FIGS. 16a to 16d, the horizontal axis represents skew, and the vertical axis represents the voltage of the first voltage signal (VOUT1), the second voltage signal (VOUT2), and the detection signal (DS), respectively.
[0109] Referring to FIG. 16a, the control logic (270) of FIG. 3 can determine a point where the slope of the detection signal (DS) is positive among the points where the first voltage signal (VOU1) and the second voltage signal (VOUT2) intersect according to the selected strobe signal as a lock point (LP11), and can determine the skew at the lock point (LP11) as a target setup margin and a target hold margin.
[0110] Referring to FIG. 16b, the control logic (270) of FIG. 3 can determine the point where the slope of the detection signal (DS) is positive among the points where the first voltage signal (VOU1) and the second voltage signal (VOUT2) intersect according to the selected strobe signal as the lock point (LP21), and can determine the skew at the lock point (LP11) as the target setup margin and the target hold margin.
[0111] Referring to FIG. 16c, the control logic (270) of FIG. 3 can determine the points where the slope of the detection signal (DS) is positive among the points where the first voltage signal (VOU1) and the second voltage signal (VOUT2) intersect according to the selected strobe signal as lock points (LP31, LP32), and can determine the skew at the lock point (LP11) as the target setup margin and the target hold margin.
[0112] Referring to FIG. 16d, the control logic (270) of FIG. 3 can determine the point where the slope of the detection signal (DS) is positive among the points where the first voltage signal (VOU1) and the second voltage signal (VOUT2) intersect according to the selected strobe signal as the lock point (LP41), and can determine the skew at the lock point (LP11) as the target setup margin and the target hold margin.
[0113] Accordingly, the ultra-high-speed transmitter according to the embodiments of the present invention controls the skew between test data and multi-phase clock signals transmitted through a separate path from the path through which the data signal is transmitted, determines a lock point having a target setup margin and a target hold margin based on the controlled skew, and transmits the data signal using adjusted clock signals having skew at the lock point, thereby enabling stable transmission of the data signal even when the data transmission rate increases.
[0114] FIG. 17 is a block diagram showing a memory system according to embodiments of the present invention.
[0115] Referring to FIG. 17, the memory system (300) includes a memory controller (305) and a semiconductor memory device (400). The memory system (300) may further include a plurality of signal lines (330) that electrically connect the memory controller (305) and the semiconductor memory device (400).
[0116] The semiconductor memory device (400) is controlled by a memory controller (300). For example, the memory controller (300) can write data to the semiconductor memory device (400) or read data from the semiconductor memory device (400) based on a request from a host (not shown).
[0117] A plurality of signal lines (330) may include a command line, an address line, a data line, and / or a power line. The memory controller (300) may transmit a command (CMD) and an address (ADDR) to a semiconductor memory device (400) through the command line and the address line, exchange data signals (MLDAT) with the semiconductor memory device (400) through the data line, and provide power (PWR) to the semiconductor memory device (400) through the power line.
[0118] For example, the data signal (MLDAT) may be a PAM-3 multi-level signal generated / transmitted according to embodiments of the present invention. Although not illustrated, a plurality of signal lines (330) may further include a DQS line that transmits a data strobe signal.
[0119] In one embodiment, some or all of the plurality of signal lines (330) may be referred to as a channel. In this specification, the data line through which a data signal (MLDAT) is transmitted is referred to as a channel. However, the present invention is not limited thereto, and the channel may further include the command line through which a command (CMD) is transmitted and / or the address line through which an address (ADDR) is transmitted.
[0120] FIGS. 18 and 19 are block diagrams illustrating examples of a system according to embodiments of the present invention.
[0121] Referring to FIGS. 18 and 19, the memory system (301) includes a graphic processing unit (GPU, 310), a semiconductor memory device (400a), and a plurality of channels (31a, 31b, 31c).
[0122] The GPU (310) may include a plurality of transmitters (25a, 25b, 25c), a plurality of receivers (27a, 27b, 27c), and a plurality of data input / output pads (29a, 29b, 29c). The semiconductor memory device (400a) may include a plurality of transmitters (45a, 45b, 45c), a plurality of receivers (47a, 47b, 47c), and a plurality of data input / output pads (49a, 49b, 49c).
[0123] Each of the multiple transmitters (25a, 25b, 25c, 45a, 45b, 45c) can generate a PAM-3 multi-level signal. Each of the multiple receivers (27a, 27b, 27c, 47a, 47b, 47c) can receive the PAM-3 multi-level signal. The multiple transmitters (25a, 25b, 25c, 45a, 45b, 45c) and the multiple receivers (27a, 27b, 27c, 47a, 47b, 47c) can transmit the PAM-3 multi-level signal through multiple channels (31a, 31b, 31c).
[0124] Each of the multiple transmitters (25a, 25b, 25c, 45a, 45b, 45c) may employ the transmitter (100) of FIG. 3.
[0125] Each of the multiple data input / output pads (29a, 29b, 29c, 49a, 49b, 49c) can be connected to one of the multiple transmitters (25a, 25b, 25c, 45a, 45b, 45c) and one of the multiple receivers (27a, 27b, 27c, 47a, 47b, 47c).
[0126] Multiple channels (31a, 31b, 31c) can connect the GPU (310) and the semiconductor memory device (400a). Each of the multiple channels (31a, 31b, 31c) can be connected to one of the multiple transmitters (25a, 25b, 25c) and one of the multiple receivers (27a, 27b, 27c) through one of the multiple data input / output pads (29a, 29b, 29c), and can be connected to one of the multiple transmitters (45a, 45b, 45c) and one of the multiple receivers (47a, 47b, 47c) through one of the multiple data input / output pads (49a, 49b, 49c). The multi-level signal can be transmitted through each of the multiple channels (31a, 31b, 31c).
[0127] FIG. 18 illustrates the operation of transmitting data from a GPU (310) to a semiconductor memory device (400a). For example, a transmitter (25a) generates an output data signal (DS11), which is a PAM-3 multi-level signal, based on input data (DAT11), and the output data signal (DS11) is transmitted from the GPU (310) to the semiconductor memory device (400a) through a channel (31a), and a receiver (47a) receives the output data signal (DS11) and can obtain target data (ODAT11) corresponding to the input data (DAT11).
[0128] Similarly, the transmitter (25b) generates an output data signal (DS21), which is a PAM-3 multi-level signal, based on the input data (DAT21), and the output data signal (DS21) is transmitted to a semiconductor memory device (400a) through the channel (31b), and the receiver (47b) can receive the output data signal (DS21) and obtain target data (ODAT21) corresponding to the input data (DAT21).
[0129] The transmitter (25c) generates an output data signal (DSN1), which is a PAM-3 multi-level signal, based on input data (DATN1), and the output data signal (DSN1) is transmitted to a semiconductor memory device (400a) through a channel (31c), and the receiver (47c) receives the output data signal (DSN1) and can obtain target data (ODATN1) corresponding to the input data (DATN1). For example, the input data (DAT11, DAT21, DATN1) may be write data written to the semiconductor memory device (400a).
[0130] FIG. 19 illustrates the operation of transmitting data from a semiconductor memory device (400a) to a GPU (310). For example, a transmitter (45a) generates an output data signal (DS12), which is a PAM-3 multi-level signal, based on input data (DAT12), and the output data signal (DS12) is transmitted from the semiconductor memory device (400a) to the GPU (310) through a channel (31a), and a receiver (27a) receives the output data signal (DS12) and can obtain target data (ODAT12) corresponding to the input data (DAT12).
[0131] Similarly, the transmitter (45b) generates an output data signal (DS22), which is a PAM-3 multi-level signal, based on the input data (DAT22), and the output data signal (DS22) is transmitted to the GPU (310) via channel (31b), and the receiver (27b) can receive the output data signal (DS22) to obtain target data (ODAT22) corresponding to the input data (DAT22). The transmitter (45c) generates an output data signal (DSN2), which is a PAM-3 multi-level signal, based on the input data (DATN2), and the output data signal (DSN2) is transmitted to the GPU (310) via channel (31c), and the receiver (27c) can receive the output data signal (DSN2) to obtain data (ODATN2) corresponding to the input data (DATN2). For example, input data (DAT12, DAT22, DATN2) may be read data read from a semiconductor memory device (400a).
[0132] FIG. 20 is a block diagram showing an example of a semiconductor memory device included in the system of FIG. 18 and FIG. 19 according to embodiments of the present invention.
[0133] Referring to FIG. 20, a semiconductor memory device (400a) may include a control logic circuit (410), an address register (420), a bank control logic (4230), a refresh counter (445), a row address multiplexer (440), a column address latch (450), a row decoder (460), a column decoder (470), a memory cell array (510), a sense amplifier section (485), an input / output gating circuit (490), an ECC engine (590), a data input / output buffer (495), and an on-die termination (ODT) circuit (497).
[0134] For example, the semiconductor memory device (400a) may be a volatile memory device, and in particular, may be a GDDR7 (graphic double data rate 7) SDRAM (synchronous dynamic random access memory) device.
[0135] The memory cell array (510) may include first to eighth bank arrays (510a to 510h). Additionally, the row decoder (460) may include first to eighth row decoders (460a to 460h) each connected to the first to eighth bank arrays (510a to 510h), the column decoder (470) may include first to eighth column decoders (470a to 470h) each connected to the first to eighth bank arrays (510a to 510h), and the sense amplifier unit (485) may include first to eighth sense amplifiers (485a to 485h) each connected to the first to eighth bank arrays (510a to 510h).
[0136] The first to eighth bank arrays (510a to 510h), the first to eighth sense amplifiers (485a to 485h), the first to eighth column decoders (470a to 470h), and the first to eighth row decoders (460a to 460h) can each constitute the first to eighth banks. Each of the first to eighth bank arrays (510a to 510h) may include a plurality of word lines (WL), a plurality of bit lines (BTL), and a plurality of memory cells (MC) formed at the intersection of the word lines (WL) and the bit lines (BTL).
[0137] The address register (420) can receive an address (ADDR) including a bank address (BANK_ADDR), a row address (ROW_ADDR), and a column address (COL_ADDR) from a memory controller included in the GPU (310). The address register (420) can provide the received bank address (BANK_ADDR) to the bank control logic (430), provide the received row address (ROW_ADDR) to the row address multiplexer (440), and provide the received column address (COL_ADDR) to the column address latch (450).
[0138] The bank control logic (430) can generate bank control signals in response to the bank address (BANK_ADDR). In response to the bank control signals, a row decoder corresponding to the bank address (BANK_ADDR) among the first to eighth row decoders (460a to 460h) can be activated, and a column decoder corresponding to the bank address (BANK_ADDR) among the first to eighth column decoders (470a to 470h) can be activated.
[0139] The row address multiplexer (440) can receive a row address (ROW_ADDR) from the address register (420) and a refresh row address (REF_ADDR) from the refresh counter (445). The row address multiplexer (440) can optionally output the row address (ROW_ADDR) or the refresh row address (REF_ADDR) as a row address (RA). The row address (RA) output from the row address multiplexer (240) can be applied to the first to eighth bank decoders (460a to 460h), respectively.
[0140] The refresh counter (445) can sequentially increase or decrease the refresh row address (REF_ADDR) under the control of the control logic circuit (410).
[0141] Among the first to eighth bank row decoders (460a to 460h), the row decoder activated by the bank control logic (430) can decode the row address (RA) output from the row address multiplexer (440) and activate the word line corresponding to the row address. For example, the activated row decoder can apply a word line driving voltage to the word line corresponding to the row address.
[0142] The column address latch (450) receives a column address (COL_ADDR) from the address register (420) and can temporarily store the received column address (COL_ADDR). Additionally, the column address latch (450) can gradually increase the received column address (COL_ADDR) in burst mode. The column address latch (450) can apply the temporarily stored column address (COL_ADDR) or the gradually increased column address (COL_ADDR') to the first to eighth column decoders (470a to 470h), respectively.
[0143] Among the first to eighth column decoders (470a to 470h), the column decoder activated by the bank control logic (430) can activate a sense amplifier corresponding to the bank address (BANK_ADDR) and the column address (COL_ADDR) through the corresponding input / output gating circuit (490).
[0144] The input / output gating circuit (490) may include input data mask logic, read data latches for storing data output from the first to eighth bank arrays (510a to 510h), and write drivers for writing data to the first to eighth bank arrays (510a to 510h), together with circuits for gating input / output data.
[0145] A codeword (CW) to be read from one of the first to eighth bank arrays (510a to 510h) is detected by a sense amplifier corresponding to the bank array and can be stored in the read data latches. The codeword (CW) stored in the read data latches can be provided to the memory controller through a data input / output buffer (495) after ECC decoding is performed by the ECC engine (590).
[0146] Multi-level data (MLDAT) to be written to one of the first to eighth bank arrays (510a to 510h) is provided to an ECC engine (590), the ECC engine (590) generates parity bits based on the multi-level data (MLDAT), and provides a codeword (CW) including the data (DQ) and the parity bits to an input / output gating circuit (490), and the input / output gating circuit (290) can write the codeword (CW) to a subpage of the one bank array through the write drivers.
[0147] The data input / output buffer (495) can provide multi-level data (MLDAT) to the ECC engine (590) during a write operation and provide multi-level data (MLDAT) provided from the ECC engine (590) to the memory controller during a read operation.
[0148] The ECC engine (590) can perform ECC encoding and ECC decoding for multi-level data (MLDAT) under the control of the control logic circuit (410).
[0149] The control logic circuit (410) can control the operation of the semiconductor memory device (400a). For example, the control logic circuit (410) can generate control signals to cause the semiconductor memory device (410a) to perform a write operation or a read operation. The control logic circuit (410) may include a command decoder (411) that decodes a command (CMD) received from the memory controller and a mode register (412) for setting the operation mode of the semiconductor memory device (200).
[0150] For example, the command decoder (411) can decode the write enable signal, the row address strobe signal, the column address strobe signal, the chip select signal, etc., to generate the control signals corresponding to the command (CMD).
[0151] The ODT circuit (497) can be connected to the data input / output pad (499) and the data input / output buffer (495). When the ODT circuit (497) is activated, an ODT can be performed. When the ODT is performed, the fidelity of the transmitted and received signals can be improved by suppressing the reflection of the signal through impedance matching.
[0152] FIG. 21 shows a first bank array in the semiconductor memory device of FIG. 20 according to embodiments of the present invention.
[0153] Referring to FIG. 21, the first bank array (510a) includes a plurality of word lines (WL0 to WLm-1, where m is an even integer greater than or equal to 2), a plurality of bit lines (BL0 to BLn-1, where n is an even integer greater than or equal to 2), and a plurality of memory cells (MCs) placed at the intersection between the word lines (WL0 to WLm-1) and the bit lines (BL0 to BLn-1).
[0154] Bit lines (BL0~BLn-1) can be extended in the first direction (D1) and word lines (WL0~WLm-1) can be extended in the second direction (D2).
[0155] Each memory cell (MC) has a DRAM cell structure. Also, it can be seen that the arrangement of memory cells (MCs) connected to each even word line (WL0) and the odd word line (WL1) is different from each other.
[0156] However, the embodiments of the present invention are not limited thereto. In the embodiments, the arrangement of memory cells (MCs) connected to each even word line (WL0) and the odd word line (WL1) may be identical to each other.
[0157] FIG. 22 is a block diagram showing a storage system according to embodiments of the present invention.
[0158] Referring to FIG. 22, the storage system (600) may include a host (610) and a storage device (650) connected via a link (605). FIG. 22 illustrates only the components for explaining embodiments of the present invention. Depending on their respective configurations, the host (610) and the storage device (650) may further include various components not illustrated in FIG. 22.
[0159] The host (610) can store data in the storage device (650) or read data stored in the storage device (650). The host (610) can communicate with the storage device (650) through the first port (PT1). For example, the first port (PT1) may be a physical port based on the PCIe protocol. However, the scope of the invention is not limited thereto.
[0160] In the following, to briefly explain the technical features of the present invention, it is assumed that the host (610) and the storage device (650) communicate with each other through a PCIe protocol-based physical port (e.g., first and second ports (PT1, PT2)).
[0161] The host (610) may include a processor (620), a host memory buffer (HMB, 630), and a root complex (640).
[0162] The processor (620) may be referred to as a host processor and may include a central processing unit (CPU) core (621) and a cache (625) assigned to the CPU core (621).
[0163] The host memory buffer (630) may include a transmission queue (SQ, 631) and a completion queue (CQ, 633). The transmission queue (631) may store commands to be provided to the storage device (650). The completion queue (633) may store completion information for operations completed in the storage device (650).
[0164] The root complex (640) can be connected to the processor (620) and the host memory buffer (630).
[0165] The storage device (650) may include a storage controller (660) and at least one non-volatile memory device (680).
[0166] At least one non-volatile memory device (680) may operate under the control of the storage controller (660). At least one non-volatile memory device (680) may be a NAND flash memory device, but the scope of the invention is not limited thereto.
[0167] The storage controller (660) may include a command parser (665) and a direct memory access (DMA) engine (670).
[0168] The command analyzer (665) analyzes the command provided from the host (610) and provides the analyzed command to the non-volatile memory device (680), and the DMA engine (670) can control memory access to the non-volatile memory device (680) or the host memory buffer (630).
[0169] In an embodiment, the host (610) and the storage device (650) can be connected via a PCIe link (605) according to the PCIe Express (PCIe) standard.
[0170] FIGS. 23 and FIGS. 24 are drawings illustrating an embodiment of an interconnect architecture applied to a storage system according to embodiments of the present invention.
[0171] Referring to FIGS. 23 and 24, an embodiment of a layered protocol stack is illustrated. The layered protocol stack includes any form of layered communication stack, such as a Quick Path Interconnect (QPI) stack, a PCIe stack, a next-generation high-performance computing interconnect stack, or other layered stacks. Embodiments of the present invention are described based on a PCIe stack, but the same concepts may be applied to other interconnect stacks. In one embodiment, the protocol stack is a PCIe protocol stack comprising a transaction layer, a link layer, and a physical layer. The representation as a communication protocol stack may also be referred to as a module or interface implementing / containing the protocol stack.
[0172] PCIe uses packets to communicate information between devices (or components). To transfer information from a sending component to a receiving component, packets can be formed at the transaction layer and the data link layer. As transmitted packets flow through other layers, they are expanded with additional information necessary to process them at those layers. At the receiving end, a reverse process takes place, converting the packets from their physical layer representation to a data link layer representation, and finally (in the case of transaction layer packets) into a form that can be processed by the transaction layer of the receiving device.
[0173] In one embodiment, the transaction layer provides an interface between the device's processing core and interconnect architecture, such as the data link layer and the physical layer. In this regard, the primary responsibility of the transaction layer is the assembly and disassembly of packets (i.e., transaction layer packets or TLPs). The transaction layer typically manages credit-based flow control for TLPs. PCIe implements split transactions, that is, transactions where the request and response are separated by time, allowing the link to return other traffic while the target device collects data for the response.
[0174] In one embodiment, the transaction layer assembles the packet headers / payloads. The format for the current packet headers / payloads can be found in the PCIe specifications on the PCIe specification website.
[0175] The link layer, also referred to as the data link layer, serves as an intermediate stage between the transaction layer and the physical layer. In one embodiment, the responsibility of the data link layer is to provide a reliable mechanism for exchanging transaction layer packets (TLPs) between two components (610, 650) over the link. One side of the data link layer accepts the TLPs assembled by the transaction layer, applies a packet sequence identifier, i.e., an identification number or packet number, calculates and applies an error detection code, i.e., an LCRC, and submits the modified TLPs to the physical layer for transmission to an external device across the physical layer.
[0176] In one embodiment, the physical layer includes a logical subblock and an electrical subblock to physically transmit packets to an external device. Here, the logical subblock is responsible for the "digital" functions of the physical layer. In this regard, the logical subblock includes a transmission section that prepares transmission information for transmission by the physical subblock, and a reception section that identifies and prepares received information before transmitting it to the link layer.
[0177] The physical block includes a transmitter (TX) and a receiver (RX). The transmitter (TX) may employ the transmitter (100) of FIG. 3 according to embodiments of the present invention. The transmitter receives symbols from a logic subblock, serializes them, and transmits them to an external device. The receiver receives serialized symbols from the external device and converts the received signals into a bit stream. The bit stream is deserialized and supplied to the logic subblock. In one embodiment, an 8b / 10b transmission code is used, and 10-bit symbols are transmitted / received. Here, specific symbols are used to frame packets into frames. In addition, in one example, the receiver also provides a symbol clock recovered from an incoming serial stream.
[0178] As described above, although the transaction layer, link layer, and physical layer are discussed with respect to specific embodiments of the PCIe protocol stack, the layered protocol stack is not limited thereto. Depending on the embodiments, any layered protocol may be included or implemented.
[0179] The PCIe interface technology has evolved through the 1st generation (GEN1), 2nd generation (GEN2), 3rd generation (GEN3), 4th generation (GEN4), and 5th generation (GEN5). For example, the 1st generation supports a data rate of 2.5 Gb / s, the 2nd generation supports 5 Gb / s, and the 3rd generation supports 8 Gb / s; the 4th generation supports a data rate of 16 Gb / s; and the 5th generation supports a data rate of 32 Gb / s.
[0180] In the future, 6th generation PCIe interfaces will support a data rate of 64 Gb / s, and accordingly, the jitter of the data signal becomes very sensitive to process, voltage, and temperature. Since the transmitter (TX) employs the transmitter (100) of FIG. 3 according to embodiments of the present invention, the setup margin and hold margin of the test data are adjusted in real time based on the test data and multiphase clock signals transmitted through a path separate from the path through which the data signal is transmitted, and a lock point having a target setup margin and a target hold margin is determined based on the adjusted setup margin and hold margin, and by transmitting a data signal having a delay at the lock point, the data signal can be transmitted stably even if the data transmission rate increases.
[0181] FIG. 25 is a block diagram showing a storage system according to embodiments of the present invention.
[0182] Referring to FIG. 25, the storage system (700) may include a universal flash storage (UFS) host (710) and a UFS storage device (740).
[0183] The UFS host (710) may include a storage interface circuit (720), and the UFS storage device (740) may include a host interface circuit (750).
[0184] In FIG. 25, the storage interface circuit (720) and the host interface circuit (750) may be referred to as the first interface circuit and the second interface circuit, respectively, for convenience of explanation, and may include M-PHY and Unipro, which are interface protocols proposed by the MIPI (Mobile Industry Processor Interface) Alliance. The first interface circuit (720) may communicate according to a defined protocol (i.e., the UFS protocol). To this end, the physical layer (M-PHY) of the first interface circuit (720) may include output terminals (DOUT_t, DOUT_c), input terminals (DIN_t, DIN_c), and a clock terminal (REF_CLK). The physical layer (M-PHY) of the first interface circuit (720) may include a transmitter (731) and a receiver (732).
[0185] The physical layer (M-PHY) of the first interface circuit (720) can transmit signals to the second interface circuit (750) through output terminals (DOUT_t, DOUT_c). The output terminals (DOUT_t, DOUT_c) can be connected to a transmitter (731) to form a transmission channel (M-TX) of the first interface circuit (720). For example, the signals transmitted through the output terminals (DOUT_t, DOUT_c) can form a pair of differential signals. That is, the signal transmitted through the output terminal (DOUT_c) can be complementary to the signal transmitted through the output terminal (DOUT_t).
[0186] The physical layer (M-PHY) of the first interface circuit (720) can receive signals from the second interface circuit (750) through input terminals (DIN_t, DIN_c). The input terminals (DIN_t, DIN_c) can be connected to a receiver (732) to form a receiving channel (M-RX) of the first interface circuit (720). For example, the signals received through the input terminals (DIN_t, DIN_c) can form a pair of differential signals. That is, the signal received through the input terminal (DIN_c) can be complementary to the signal received through the input terminal (DIN_t).
[0187] The output terminals (DOUT_t, DOUT_c) and input terminals (DIN_t, DIN_c) can be controlled to one of various states according to a predetermined convention. For example, each of the output terminals (DOUT_t, DOUT_c) and input terminals (DIN_t, DIN_c) can be controlled to one of a positive state, a negative state, a ground state, or a floating state.
[0188] When the level of the output signal of the first output terminal (DOUT_t) (e.g., voltage level) is higher than the level of the output signal of the second output terminal (DOUT_c), the output terminals (DOUT_t, DOUT_c) may be in a positive state. When the level of the output signal of the first output terminal (DOUT_t) is lower than the level of the output signal of the second output terminal (DOUT_c), the output terminals (DOUT_t, DOUT_c) may be in a negative state. When the first output terminal (DOUT_t) and the second output terminal (DOUT_c) are floating, the output terminals (DOUT_t, DOUT_c) may be in a floating state. When the levels of the first output terminal (DOUT_t) and the second output terminal (DOUT_c) are the same, the output terminals (DOUT_t, DOUT_c) may be in a ground state.
[0189] When the level of the input signal at the first input terminal (DIN_t) is higher than the level of the input signal at the second input terminal (DIN_c), the input terminals (DIN_t, DIN_c) may be in a positive state. When the level of the input signal at the first input terminal (DIN_t) is lower than the level of the input signal at the second input terminal, the input terminals (DIN_t, DIN_c) may be in a negative state. When the first input terminal (DIN_t) and the second input terminal (DIN_c) are connected to terminals in a ground state, the output terminals (DOUT_t, DOUT_c) may be in a ground state. When the first input terminal (DIN_t) and the second input terminal (DIN_c) are floating, the input terminals (DIN_t, DIN_c) may be in a floating state.
[0190] The second interface circuit (750) can communicate according to the UFS protocol. To this end, the physical layer (MPHY) of the second interface circuit (750) may include output terminals (DOUT_t, DOUT_c), input terminals (DIN_t, DIN_c), and a clock terminal (REF_CLK).
[0191] The output terminals (DOUT_t, DOUT_c) of the second interface circuit (750) may correspond to the input terminals (DIN_t, DIN_c) of the first interface circuit (720), and the input terminals (DIN_t, DIN_c) of the second interface circuit (750) may correspond to the output terminals (DOUT_t, DOUT_c) and input terminals (DIN_t, DIN_c) of the first interface circuit (720).
[0192] The physical layer (M-PHY) of the second interface circuit (750) can receive signals through input terminals (DIN_t, DIN_c) and transmit signals through output terminals (DOUT_t, DOUT_c). The physical layer (M-PHY) of the second interface circuit (750) may include a receiver (761) and a transmitter (763).
[0193] The output terminals (DOUT_t, DOUT_c) and input terminals (DIN_t, DIN_c) of the second interface circuit (750) can be controlled to a positive state, a negative state, a ground state, or a floating state in a manner similar to that described above for the first interface circuit (720).
[0194] Meanwhile, according to the MIPI M-PHY specification, the physical layer (M-PHY) of the second interface circuit (750) may include a reference clock detector (765). The reference clock detector (765) can detect a change between the idle mode and the active mode of the UFS storage device (740).
[0195] When the UFS storage device (740) is not performing any operation, the UFS storage device (740) may be in a first idle mode or a second idle mode. When the UFS storage device (740) is in the first idle mode and / or the second idle mode, the first interface circuit (720) may not transmit the reference clock (REF_CLK) to the second interface circuit (750). When the UFS storage device (740) changes from the first idle mode and / or the second idle mode to an active mode, the input terminals (DIN_t, DIN_c) of the second interface circuit (750) may change from a floating state to a negative state. And, when the UFS storage device (740) changes from the first idle mode and / or the second idle mode to an active mode, the first interface circuit (720) may resume transmitting the reference clock (REF_CLK) to the second interface circuit (750).
[0196] In an embodiment, when the UFS storage device (740) is in a second idle mode, the reference clock detector (765) can generate a trigger signal to enter the UFS storage device (740) into an active mode based on the toggling of the reference clock (REF_CLK).
[0197] Each of the transmitters (731, 763) can employ the transmitter (100) of Fig. 3.
[0198] In MIPI M-PHY gear 6, the data transmission rate is further increased, and since each of the transmitters (731, 763) employs the transmitter (100) of FIG. 3, the setup margin and hold margin of the test data are adjusted in real time based on the test data and multi-phase clock signals transmitted through a path separate from the path through which the data signal is transmitted, and a lock point having a target setup margin and a target hold margin is determined based on the adjusted setup margin and hold margin, and by transmitting a data signal having a delay at the lock point, the data signal can be transmitted stably even when the data transmission rate is increased.
[0199] FIG. 26 is an exemplary block diagram showing a semiconductor memory device according to embodiments of the present invention.
[0200] Referring to FIG. 26, a semiconductor memory device (800) may include at least one buffer die (810) and a plurality of memory dies (820-1, 820-2, ..., 820-p, where p is a natural number greater than or equal to 3) to provide a function for analyzing and rescuing soft data failure in a stacked chip structure.
[0201] Multiple memory dies (820-1, 820-2, ..., 820-p) are sequentially stacked on top of a buffer die (810) and can communicate data through multiple through silicon via (hereinafter TSV) lines.
[0202] At least one of the plurality of memory dies (820-1, 820-2, ..., 820-p) may include a cell core ECC engine (823) that generates transmission parity bits using transmission data transmitted to a cell core (821) that stores data and a buffer die (810). The cell core (821) may include a plurality of memory cells having a DRAM cell structure.
[0203] The buffer die (810) may include a via ECC engine (812) that generates error-corrected data by correcting the transmission error using transmission parity bits when a transmission error occurs in the transmission data received through the plurality of TSV lines.
[0204] The buffer die (810) may also include a transmitter (814) and a receiver (813). The transmitter (184) may include the transmitter (1000) of FIG. 3. Thus, the transmitter (814) can generate test data in parallel with the data provided from the via ECC engine (814), and based on the test data, adjust the delay amount of the data to transmit an output data signal having a target setup margin and a target hold margin to an external memory controller.
[0205] The semiconductor memory device (800) may be a stacked chip type memory device or a stacked memory device that communicates the data and control signals through the TSV lines. The TSV lines may also be referred to as through-silicon electrodes.
[0206] The cell core ECC engine (822) can also perform error correction on the data output from the memory die (820-p) before the transmitted data is transmitted.
[0207] Transmission errors occurring in the transmitted data may be generated due to noise occurring in the TSV lines. Data failures generated due to noise occurring in the TSV lines may be soft data failures, as they are distinct from data failures caused by the memory die itself. Since such soft data failures are caused by transmission path failures, they can be detected and recovered by implementing ECC operations.
[0208] A data TSV line group (832) formed on a single memory die (820-p) may be composed of TSV lines (L1, L2~Lp), and a parity TSV line group (834) may be composed of TSV lines (L10~Lq). The TSV lines (L1, L2~Lp) of the data TSV line group (832) and the TSV lines (L10~Lq) of the parity TSV line group (834) may be connected to microbumps (MCBs) correspondingly formed between a plurality of memory dies (820-1~820-p).
[0209] The semiconductor memory device (800) may have a 3D chip structure or a 2.5D chip structure to communicate with an external memory controller via a data bus (B10). The buffer die (810) may be connected to the memory controller via the data bus (B10).
[0210] The cell core ECC engine (823) outputs transmission data through the data TSV line group (832). Additionally, the cell core ECC engine (823) outputs transmission parity bits through the parity TSV line group (834). The output transmission data may be data that has been error-corrected by the cell core ECC engine (832).
[0211] The via ECC engine (812) checks whether a transmission error has occurred in the transmission data received through the data TSV line group (832) using the transmission parity bits received through the parity TSV line group (834). If a transmission error occurs, the via ECC engine (812) corrects the transmission error in the transmission data using the transmission parity bits. If the number of bits of the transmission error cannot be corrected, the via ECC engine (812) can output information indicating that a data error has occurred.
[0212] FIG. 27 is a structural diagram showing an example of a semiconductor package including a stacked memory device according to embodiments of the present invention.
[0213] Referring to FIG. 27, the semiconductor package (900) may include one or more stacked memory devices (910) and a graphics processor (GPU, 920), and the graphics processor (920) may include a memory controller (925).
[0214] The stacked memory device (910) and the graphics processor (920) are mounted on an interposer (930), and the interposer (930) on which the stacked memory device (910) and the graphics processor (920) are mounted can be mounted on a package substrate (940). The package substrate (940) can be mounted on a solder ball (950).
[0215] The stacked memory device (910) can be implemented in various forms, and according to one embodiment, the stacked memory device (910) may be a memory device in the form of High Bandwidth Memory (HBM) in which a plurality of layers are stacked. Accordingly, the stacked memory device (910) includes a buffer die and a plurality of memory dies, and the plurality of memory dies may each include a cell core and a cell core ECC engine, and the buffer die may include a via ECC engine, a transmitter, and a receiver.
[0216] A plurality of stacked memory devices (910) may be mounted on the interposer (930), and a graphics processor (920) may communicate with the plurality of stacked memory devices (910). As an example, each of the stacked memory devices (910) and the graphics processor (920) may include a physical region, and communication may be performed between the stacked memory devices (910) and the graphics processor (920) through a physical (PHY) region. Industrial applicability
[0217] The present invention can be usefully applied to a memory device including an ultra-high-speed transmitter and any host device.
[0218] Although the present invention has been described above with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as set forth in the following claims.
Claims
Claim 1 A high-speed transmitter comprising: a pattern generator that generates test data having a fixed pattern; a data generator that generates a retimed data signal and a retimed test data having a setup margin and a hold margin, respectively, for the setup margin and hold margin of the input data signal and the test data, by adjusting the delay amount of each of the input data signal and the test data based on adjusted clock signals; a serializer that serializes the retimed data signal based on multiphase clock signals to generate a serialized data signal; a transmission driver that generates an output data signal based on the serialized data and transmits the output data signal through a channel; and a feedback circuit that detects the setup margin and the hold margin of the retimed test data through a path separate from the path of the retimed data signal, and generates the adjusted clock signals by adjusting the delay amount of the multiphase clock signals based on the detected setup margin and the detected hold margin. Claim 2 The ultra-high-speed transmitter according to claim 1, wherein the feedback circuit comprises: a first dummy multiplexer that sequentially selects a first portion of test bits of the retimed test data based on the multiphase clock signals and provides it as a first output signal; a second dummy multiplexer that sequentially selects a second portion of test bits of the retimed test data different from the first portion and provides it as a second output signal based on the multiphase clock signals; a sensing circuit that detects the setup margin and the hold margin of the retimed test data based on the first output signal, the second output signal, and the strobe signal, and generates a sensing signal indicating the detected setup margin and the detected hold margin; a strobe signal selector that provides one selected from the test bits of the retimed test data as the strobe signal; a control logic that generates a delay control signal based on the sensing signal; and a clock timing controller that adjusts the delay amount of the multiphase clock signals based on the delay control signal to generate the adjusted clock signals. Claim 3 In paragraph 2, the ultra-high-speed transmitter is characterized in that the first dummy multiplexer sequentially selects the first portion of the test bits based on a first clock signal and a second clock signal having a phase difference of 90 degrees among the multi-phase clock signals, and the second dummy multiplexer sequentially selects the second portion of the test bits based on the first clock signal and the second clock signal among the multi-phase clock signals. Claim 4 In paragraph 2, the sensing circuit comprises: a first exclusive OR gate that performs an exclusive OR operation on the first output signal and the strobe signal to provide a third output signal; a second exclusive OR gate that performs an exclusive OR operation on the second output signal and the strobe signal to provide a fourth output signal; a first integrator that integrates the third output signal to output a first voltage signal which is an analog signal; and a second integrator that integrates the fourth output signal to output a second voltage signal which is an analog signal. An ultra-high-speed transmitter comprising a comparator that subtracts the second voltage signal from the first voltage signal to output the detection signal, wherein the third output signal has a first duty representing a first skew between the retimed test data and the multiphase clock signals, and the fourth output signal has a second duty representing a second skew between the retimed test data and the multiphase clock signals, and wherein the control logic determines at least one point among the points where the first voltage signal and the second voltage signal intersect based on the detection signal, wherein the slope of the detection signal is positive, as a lock point, wherein at the lock point the retimed data signal has a target setup margin and a target hold margin, and wherein the lock point varies based on the strobe signal selected by the strobe signal selector. Claim 5 An ultra-high-speed transmitter according to claim 4, wherein, in response to the detection signal indicating that the first voltage signal is greater than the second voltage signal, the control logic generates the delay control signal to reduce the delay amount, and the clock timing regulator generates the adjusted clock signals by reducing the delay amount of the multi-phase clock signals in response to the delay control signal. Claim 6 An ultra-high-speed transmitter according to claim 4, wherein, in response to the detection signal indicating that the first voltage signal is smaller than the second voltage signal, the control logic generates the delay control signal to increase the delay amount, and the clock timing regulator generates the adjusted clock signals by increasing the delay amount of the multi-phase clock signals in response to the delay control signal. Claim 7 An ultra-high-speed transmitter according to claim 4, wherein the control logic adjusts the delay control signal such that the delay amount is adjusted until the detection signal indicates that the first voltage signal is identical to the second voltage signal, and locks the delay control signal in response to the first voltage signal being identical to the second voltage signal. Claim 8 The ultra-high-speed transmitter according to claim 1, wherein the data generator comprises: a first shift register block including a plurality of first shift registers that generate the retimed data signal by delaying the data bits of the input data signal in parallel based on a first adjusted clock signal and a second adjusted clock signal having a phase difference of 180 degrees among the adjusted clock signals; and a second shift register block including a plurality of second shift registers that generate the retimed test data by delaying the test bits of the test data in parallel based on the first adjusted clock signal and the second adjusted clock signal. Claim 9 A storage device comprising: a processor that generates an input data signal; and an ultra-high-speed transmitter that generates test data and, based on the test data, adjusts the delay amount of the input data signal to have a target setup margin and a target hold margin for the setup margin and hold margin of the input data, and transmits the output data signal to a host via a link, wherein the ultra-high-speed transmitter includes: a pattern generator that generates the test data having a certain pattern; a data generator that, based on adjusted clock signals, adjusts the delay amount of each of the input data signal and the test data to generate a retimed data signal and a retimed test data having a setup margin and a adjusted hold margin, respectively, for the setup margin and hold margin of each of the input data signal and the test data; a serializer that serializes the retimed data signal based on multi-phase clock signals to generate a serialized data signal; and a transmission driver that generates the output data signal based on the serialized data and transmits the output data signal through a channel. A storage device comprising a feedback circuit that detects the setup margin and the hold margin of the retimed test data through a path separate from the path of the retimed data signal, and generates the adjusted clock signals by adjusting the delay amount of the multiphase clock signals based on the detected setup margin and the detected hold margin. Claim 10 In claim 9, the storage device communicates with the host based on the PCI (Peripheral Component Interconnect) Express protocol or the storage device communicates with the host based on the Universal Flash Storage (UFS) protocol, and the transmitter is included in the physical layer of the interface circuit of the storage device, and the physical layer is a Mobile Industry Processor Interface (MIPI) M-PHY.
Citation Information
Patent Citations
Circuit for generating accurate clock phase signals for a high-speed serializer / deserializere
KR1020160126086A
Transmitter and transmitter / receiver
US20090052599A1
Semiconductor device having serializer converting parallel data into serial data to output serial data from output buffer circuit
US20130082737A1
Memory device and divided clock correction method thereof
US20200013441A1
