Low power consumption high performance clock path architecture

By employing a clock path architecture with multiplexers and average duty cycle sensors in a high-speed wired transmitter, combined with a phase detector and independent clock design for the data path, the problems of clock alignment and power consumption optimization are solved, achieving low-power, high-performance signal transmission.

CN119030680APending Publication Date: 2024-11-26ALTERA CORP
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202311824242.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2023-05-26
Filing Date
2023-12-27
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

High-speed wired transmitters face challenges in optimizing performance and power consumption, and existing technologies struggle to achieve high-precision clock alignment, reduce power consumption, and minimize random mismatches.

Method used

A clock path architecture using a multiplexer and an average duty cycle sensor, combined with an independent clock design for the phase detector and data path, achieves precise clock signal alignment and power consumption optimization through a phase sampler and duty cycle control circuit.

Benefits of technology

It achieves low-power, high-performance clock signal transmission, reduces sensor load and switching jitter, improves signal-to-noise ratio and distortion ratio (SNDR), adapts to process voltage and temperature changes, and reduces the risk of data corruption.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119030680A_ABST
    Figure CN119030680A_ABST
Patent Text Reader

Abstract

A data transmitter having a phase detector, an average duty cycle sensor, and a phase sampler is used to optimize a clock / data path. Phase information and duty cycle information are provided to a digital control section to adjust timing in the data path and the clock path, respectively. The phase detector reads a skew between the data within the driver and negative and positive phase clock signals. An optimal pulse width increment is determined by the target duty cycle sensor. Using the measured average duty ratio sensor, the digital control unit calculates the duty ratio error of the target value required inside the driver. The phase sampler has a multiplexer that routes the clock signal to a phase sensor that determines a phase error based on, for example, a rising edge-to-rising edge comparison and a falling edge-to-falling edge comparison. Furthermore, it comprises a duty cycle sensor for each clock phase.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Generally speaking, the present application relates to high-speed wired transmitters. Background Art

[0002] For example, high-speed wired transmitters have data rates exceeding 50Gbps and often include circuits for generating high-precision clock signals. Such transmitters can be used in a variety of applications, including intra-chip and chip-to-chip transmission. In one approach, the transmitter is compatible with Ethernet networking technology. However, there are various challenges in optimizing performance and power consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0003] Embodiments of the present disclosure will be more fully understood according to the detailed description given below and the accompanying drawings according to various embodiments of the present disclosure, however, these embodiments should not be construed as limiting the present disclosure to specific embodiments but are only for explanation and understanding.

[0004] Figure 1 Depicted are tables of transmitter performance parameters in accordance with various embodiments.

[0005] Figure 2 An example clock sensing circuit is depicted that uses a different phase sensor for each clock signal for phase alignment of a set of four clock signals in accordance with various embodiments.

[0006] Figure 3 An example clock sensing circuit is depicted that uses a single multiplexed sensor for phase alignment of a set of four clock signals in accordance with various embodiments.

[0007] Figure 4 Depicted are example circuits for adjusting a gating duty cycle of a clock signal using a duty cycle sensor in accordance with various embodiments.

[0008] Figure 5 Depicted are example data alignment circuits for clock signals according to various embodiments.

[0009] Figure 6 Depicted are example data alignment circuits for clock signals using separate data and driver clocks according to various embodiments.

[0010] Fig. 7A An example D-type flip-flop 700 for detecting a phase difference between a data signal data2 and a clock signal clk4 is depicted in accordance with various embodiments.

[0011] Figure 7B Depicted are various embodiments of Fig. 7A Consistent example timing diagram.

[0012] Fig. 8ADepicted are time-interleaved complementary metal oxide semiconductor (CMOS) stages according to various embodiments.

[0013] Figure 8B Depicted are various embodiments of Fig. 8A Consistent example timing diagram.

[0014] Fig. 9 An example transmitter 900 is depicted in accordance with various embodiments.

[0015] Fig. 10A Depicted are various embodiments of Fig. 9 Examples of consistent positive phase (PPH) and negative phase (NPH) clock signals.

[0016] Fig. 10B Depicted are various embodiments, Fig. 9 An example of a four-level pulse amplitude modulation (PAM4) signal output by driver 951.

[0017] Fig.11A1 Depicted are various embodiments, Fig. 9 An example block diagram of a phase sampler 940.

[0018] Fig.11A2 Depicted are various embodiments of Fig.11A1 An example timing diagram associated with the P2N sensor 1021.

[0019] Fig. 11B Depicted are various embodiments, Fig.11A1 An example implementation of the multiplexer 1013 is provided.

[0020] Fig. 12A Depicted are various embodiments of Fig. 11B An example timing diagram consistent with the multiplexer segment 1110.

[0021] Fig. 12B Depicted are various embodiments of Fig. 11B An example timing diagram consistent with the multiplexer portion 1120 is shown.

[0022] Fig. 12C Depicted are various embodiments of Fig. 11B An example timing diagram consistent with the multiplexer portion 1130 is shown.

[0023] Fig.12D Depicted are various embodiments of Fig. 11B An example timing diagram consistent with the multiplexer portion 1140 is shown.

[0024] Fig.13A Depicted are various embodiments, Fig. 9 An example implementation of a target duty cycle sensor 955 is provided.

[0025] Fig. 13B Depicted are various embodiments, Fig. 9 An example implementation of an average duty cycle sensor 952 is provided.

[0026] Fig.14A Depicted are various embodiments, Fig. 13B An example graph of the target and sensed output voltage of the positive phase (PPH) average duty cycle sensor 952 relative to incremental pulse width.

[0027] Fig. 14B Depicted are various embodiments, Fig. 13B An example graph of the target and sensed output voltage of the negative phase (NPH) average duty cycle sensor 952 relative to incremental pulse width.

[0028] Fig.15 Depicted are various embodiments, Fig. 9 An example implementation of a phase detector 950 is provided.

[0029] Fig.16 Depicted are various embodiments of Fig.15 An example timing diagram of the phase detector 950 is shown in FIG.

[0030] Fig.17 Examples of components that may be present in computing system 1750 for implementing the techniques (eg, operations, processes, methods, and method techniques) described herein are shown. DETAILED DESCRIPTION

[0031] As mentioned at the outset, there are various challenges in optimizing the performance and power consumption of high-speed wired transmitters.

[0032] High-speed cable transmitters typically include circuitry for generating high-precision strobe signals (also known as clock signals or clocks). In addition, output drivers may be provided that typically employ some degree of interleaving (4:1 is a common number). To maximize transmitter performance in terms of metrics such as jitter and signal-to-noise and distortion ratio (SNDR), accurate clock alignment is critical. To achieve good clock alignment, delay elements may be used in different branches of the clock generation system, with tuning algorithms controlling these delays. The algorithm may make decisions based on readings from various sensors that assess the accuracy of the generated clock.

[0033] One possible way to implement the transmitter involves using separate phase sensor circuits to measure the phase mismatch between a pair of clocks on adjacent paths. However, this results in random mismatches between different sensor circuits. See Figure 2 As an alternative, one phase sensor with a multiplexer is used to measure the phase mismatch between the clocks for each possible pair of clocks. However, this requires a large multiplexer and may result in power loss and switching glitches. For more information, see Figure 3 .

[0034] Another possible approach involves using a duty cycle sensor to track the strobe signal inside the driver and adjust the clock generation to ensure strobe quality. However, adjusting the clock duty cycle to a set value does not account for process voltage temperature (PVT) variations of the driving device. In addition, if the driving device is of different types, such as p-type or n-type transistors, common-mode noise may be introduced. For more information, see Figure 4-Figure 6 .

[0035] The solutions provided herein address the above challenges and others. In one approach, a clock path architecture for an interleaved transmitter includes a multiplexer (mux.) that allows sensors (e.g., phase sampler circuits) to be switched between multiple pairs of clock signals. Thus, random mismatches of the sensors do not affect the readings. The multiplexer can be designed to introduce relatively small random mismatches, loads on the clock tree, and jitter during switching. In addition, a duty cycle control circuit (e.g., an average duty cycle sensor) can provide optimal or other preferred or desired duty cycle tracking for PVT variations. Setting the duty cycle to a near-optimal level improves the jitter and SNDR of the transmitter. In addition, the clock-to-data design can achieve sub-unit interval (UI) alignment of the data stream with the driver clock. This alignment avoids data corruption caused by PVT variations at high data rates. Different clocks can also be used in the driver and data paths, significantly reducing the power consumption of the clock generated in the clock multiplexer and entering the data path.

[0036] These and other features will become more apparent in light of the following discussion.

[0037] Figure 1A table of transmitter performance parameters according to various embodiments is depicted. These parameters are from the Optical Internetworking Forum (OIF), Common Electrical I / O (CEI)-Electrical and Jitter Interoperability agreements for 6G+bps, 11G+bps, 25G+bps, 56G+bps and 112G+bps I / O, IA # OIF-CEI-05.0, May 5, 2022. More specifically, these parameters are in Section 27.3.1 of OIF CEI-112G-LR-PAM4 Long Reach Interface. The first column describes the name of the parameter, the second column describes the associated pass limit, and the third column describes the performance impact. Figure 9-Figure 16 The transmitter circuit (transmitter or TX) described in this article has been tested and verified to exceed these performance parameters.

[0038] The transmitter circuit includes sensor circuits for clock distribution that enable low power consumption and high performance transmission. These circuits include a MUX circuit in a phase sampler for clock sensing based on a bootstrap switch, an average duty cycle sensor that generates an optimal or otherwise preferred or desired duty cycle target and measures the average duty cycle of the clock of the TX driver, and a phase detector that measures the clock to data phase difference. For example, these circuits can reduce power consumption and improve performance of jitter and SNDR parameters. For example, the following example combines Figure 9-Figure 16 The details of the circuit are further described.

[0039] The power consumption of the different circuits of the transmitter is broken down as follows: driver 35mW, clock path 35mW, data path 23mW, LDO (low dropout regulator) 1mW and PLL (phase-locked loop) 10mW. The total power consumption at 116Gbps PAM4 is 104mW, resulting in a very high energy efficiency of 0.9pJ / bit. PAM4 stands for four-level pulse amplitude modulation.

[0040] Figure 2An example clock sensing circuit according to various embodiments is depicted, which uses a different phase sensor for each clock signal for phase alignment of a set of four clock signals. The circuit receives four clock signals clk1, clk2, clk3, and clk4 on paths 201, 202, 203, and 204, respectively. A first sensor 210 is connected to paths 201 and 202 to receive clk1 and clk2, respectively, and determine the phase difference between them. A second sensor 211 is connected to paths 202 and 203 to receive clk2 and clk3, respectively, and determine the phase difference between them. A third sensor 212 is connected to paths 203 and 204 to receive clk3 and clk4, respectively, and determine the phase difference between them. A fourth sensor 213 is connected to paths 201 and 204 to receive clk1 and clk4, respectively, and determine the phase difference between them.

[0041] This is an example implementation of clock sensing that involves using a separate phase sensor circuit to measure the phase mismatch between a pair of clocks on adjacent paths, as mentioned at the beginning. Four clock signals are provided as an example, and a separate sensor circuit is connected to each pair of consecutive clock signals. However, one disadvantage of this approach is the random mismatch between different sensor circuits. As alignment targets become more stringent (e.g., below 100fSec in 100Gbps transmission), sensor circuits with sufficiently low random mismatch become too complex or large, thereby overloading the clock distribution network.

[0042] Figure 3 An example clock sensing circuit is depicted that uses a single multiplexed sensor for phase alignment of a set of four clock signals in accordance with various embodiments. The circuit receives four clock signals clk1, clk2, clk3, and clk4 on paths 301, 302, 303, and 304, respectively. Both a first multiplexer 310 and a second multiplexer 320 are connected to each of the paths 301-304 and provide respective outputs to a sensor circuit 330. Each multiplexer provides one clock at a time so that the sensor can determine the phase difference between any combination of two clock signals.

[0043] This is an example implementation of clock sensing that involves using a phase sensor circuit that switches between phase pairs as mentioned at the beginning. Sensor circuit 330 may represent Figure 2Any of the sensor circuits 210-213. The sensor circuit 330 can be connected to different clock signal pairs using two 4:1 multiplexers. Since each of the four possible pairs uses the same sensing circuit, random mismatches of the sensor blocks will not affect the results. Instead, the random mismatch between different multiplexer shoulders becomes critical. Since the multiplexer is a relatively simple circuit, shoulder-to-shoulder random mismatch is easier to manage than the potentially complex logic of the sensor circuits. Nevertheless, for precise phase alignment, the multiplexer device size needs to be very large. Shoulder-to-shoulder random mismatch refers to the variation in gain or offset between adjacent channels or inputs.

[0044] Figure 4 An example circuit for adjusting the gate duty cycle of a clock signal using a duty cycle sensor according to various embodiments is depicted. A clock generator 410 receives a target duty cycle and a feedback duty cycle signal from a duty cycle sensor 420. The clock generator 410 outputs the clock signal to a buffer 415, which in turn provides a gate signal strb to a duty cycle sensor and driver 425. The method uses a duty cycle sensor to adjust the gate duty cycle.

[0045] High-speed distribution networks often distort the gating / clock signal. Therefore, different types of sensors can be used to track the gating signal inside the driver and adjust the clock generation to ensure the gating quality. One of the parameters that can be tracked is the duty cycle, which can be read by averaging the signal. For example, when using an N to 1 multiplexer, the duty cycle can be adjusted to 1 / N. However, the measured clock duty cycle is completed just before the clock enters the driver 425. Inside the driver, the gating pulse passes through an additional gate before sampling the data. In addition, the target duty cycle of 1 / N is not necessarily the best or otherwise preferred or desired duty cycle. For example, in a voltage mode driver, the pMOS to nMOS intensity ratio may change due to PVT changes, causing the optimal duty cycle to deviate from 1 / N. pMOS represents p-type metal oxide silicon field effect transistor (MOSFET) and nMOS represents n-type MOSFET.

[0046] Figure 5An example data alignment circuit for clock signals according to various embodiments is depicted. This involves data alignment using a relatively fast (e.g., high frequency) clock. The circuit includes a clock path 510 that outputs a clock clk to a data path 520 and a driver 525 on a path 515. The data path provides data to the driver on a data line 521. In general, the data provided to the driver must be synchronized with the driver enable (clock) signal. One way to achieve alignment is to use the driver enable signal to propagate the data in the data path so that the data obtains implicit alignment. To achieve this alignment, a latch or flip-flop based retimer is used at the last stage of the data path.

[0047] The disadvantage of this solution is the load that the data path puts on the driver clocks. These clocks directly affect the transmission jitter, so they should be of the highest possible quality (high slew rate signals). Any additional load on the driver clocks will result in huge power penalties and complicate the design in terms of reliability, power delivery, etc. Therefore, it is desirable to use different clocks in the data path. Note that the multiplexing happens inside the driver, so the data path essentially "operates" at a much lower frequency compared to the driver, i.e., with 4:1 interleaving, the maximum data path frequency is one quarter of the driver frequency. Therefore, using a high frequency clock to propagate low frequency data is not power efficient.

[0048] Figure 6 An example data alignment circuit for clock signals using separate data and driver clocks is depicted in accordance with various embodiments. The circuit includes a clock path 610 that provides a data clock to a data path 625, and a clock path 615 that provides a drive (drv) clock to the clock path 610 and to a driver 635. The data path 625 provides data to the driver on the data line 630 and to the phase detector 620.

[0049] The data clock must be aligned with the driver clock. This can be done at the phase detector 620 by measuring the phase of the data signal in the driver relative to the driver clock and delaying the data clock to achieve synchronization. This approach does not require a latch-based retimer and also relaxes the clock jitter requirements in the data clock, resulting in significant power savings.

[0050] Fig. 7A An example D-type flip-flop 700 for detecting a phase difference between a data signal data2 and a clock signal clk4 is depicted in accordance with various embodiments. One method of detecting data alignment is to sample the data signal using the rising edge of the driver clock signal using a D-type flip-flop. By considering only those samples that occur on the rising edge of the data signal and aligning the data so that half of the samples are "0" and half are "1", the rise of the data signal will be aligned with the rise of the clock signal. Figure 7B An example of such an alignment with four clock phases is shown in .

[0051] Figure 7B Depicted are various embodiments of Fig. 7A Example timing diagram for alignment. The waveform includes data2, which can transition from high to low and back to high, or from low to high and back to low at time points t0 and t4. Clk1 is high from t3 to t4, clk2 is high from t2 to t3, clk3 is high from t1 to t2, and clk4 is high from t0 to t1. From t0 to t2, the setup time (tsu) is 2UI, and from t3 to t4, the hold time (th) is 1UI. This can be reversed so that the setup time is 1UI and the hold time is 2UI. UI (or unit interval) is a pulse time or symbol duration. It can be the shortest time required to send one bit of digital data. Since the data is aligned to one of the clock edges, the granularity of the alignment is 1UI.

[0052] Combined with the above Figures 2 to 7B The described method has several disadvantages. For example, Figure 2 and Figure 3 For sensor multiplexer solutions, phase alignment mechanisms typically require very large multiplexers, either tri-state based or transmission gate based, which connect pairs of clock phases to a common sensor circuit. The size is primarily defined by the random mismatch between adjacent multiplexer shoulders. Large multiplexers can cause two problems. The first is excessive load on the fast clock, which incurs a power penalty. The second is switching events. Specifically, when the multiplexer switches from one pair of clock phases to the next, brief disturbances can be introduced on the clock line due to the switching of large capacitive loads. This has the potential to cause single error events in transmitter operation. Furthermore, such events can occur repeatedly during clock adaptation.

[0053] Regarding the aforementioned Figure 4 The average duty cycle sensor method in , adjusts the clock duty cycle to a set value without considering PVT variations of the driving device. For example, in a 116Gbps PAM4 implementation, a MOS device can easily exhibit a 150mV over-corner threshold voltage shift. Considering that the MOS device is "active" when Vgs>Vth, the gate pulse rise time is 7pSec / 750mV, which corresponds to a nearly 3pSec modulation of the active period. At a symbol rate of 58Gbps, this is almost 20% of the unit interval. In addition, the threshold voltage can easily shift another 100mV over the operating temperature range of -40 to +100C.

[0054] If the drive device is of a different type (ie P / N), another problem will arise as described below.

[0055] Fig. 8A A time-interleaved CMOS stage according to various embodiments is depicted. CMOS stage 800 includes a pMOS transistor 810 connected in series with an nMOS transistor 820. An output path 825 is connected to a midpoint 815 between the two transistors. The control gate of the pMOS transistor receives a voltage strbp_i, while the control gate of the nMOS transistor receives a voltage strbn_i, where _i represents the i-th stage.

[0056] Figure 8B Depicted are various embodiments of Fig. 8A Consistent example timing diagram. Specifically, waveforms 850 and 860 represent strbp and strbn, respectively. In this case, not only the output pulse shape is affected, but also common mode noise is introduced because different devices experience different threshold voltage (Vt) deviations in both process corners and temperature ranges. For example, the active period can be changed from a relatively short period when Vth is relatively high to a relatively long period when Vth is relatively low.

[0057] Regarding the aforementioned Figure 6 The purpose of separating the data path from the driver clock in the phase detector approach shown is to allow the data path to be as "sloppy" as possible, thus saving power. For example, the data path can even be powered by a low-voltage noisy digital supply. To achieve this, the margin for data signal jitter should be maximized. An alignment granularity of 1UI may not be optimal, as it may be necessary to place the driver clock phase exactly in the middle of the stable period of the corresponding data phase. For example, see Figure 7B , where clk2 is in the middle of data2. Therefore, it may be advantageous to align the phase in finer steps, especially as the data rate increases (and the UI time decreases).

[0058] Fig. 9 An example transmitter 900 is depicted in accordance with various embodiments. The TX and clock path architectures are discussed first. Details of the bootstrapped nMOS switch based sensor multiplexer, average duty cycle sensor, and clock to data phase detector are provided further below.

[0059] The transmitter can be conceptually divided into a clock path 960 and a data path 970. In the clock path source, an inductor-capacitor (LC) PLL 920 generates a differential 4-UI clock on path 921 for use by a quadrature (quad.) generator (gen) 922. The two clocks include a pair of signals of opposite polarity and in phase with each other, where the difference between the two signals represents the data being transmitted.

[0060] In one possible embodiment, at the highest baud rate (e.g., 116 Gbps), the clock frequency is 14.5 GHz (UI=17.2 pSec). The quadrature generator 922 can generate four phases (1 UI interval) with a duty cycle of 50% on path 923. That is, four clock signals with four corresponding phases are generated. Typically, an N-phase clock (where N≥1) refers to N separate clock signals on N respective paths (e.g., wires). With respect to 1 UI spacing, this refers to the phase difference between the clocks. The clock on path 923 is provided to the clock multiplexer 911 and the quadrature sampler 931. The quadrature sampler 931 determines the phase error in the input clock signal and, in response, provides the output on path 934 to the digital control circuit 936. In response to the output, the digital control circuit provides a control signal on path 934a to adjust the quadrature generator. The digital control unit (digitalcontrol) may include a processor 936b for executing instructions stored in the memory 936a.

[0061] Clock multiplexer 911 provides an 8-UI, 50% duty cycle clock to a low frequency parallel input, serial output circuit (LF PISO) 905 via path 910, and provides an 8-UI, 50% duty cycle clock to a high frequency PISO (HF PISO) 908 via path 912. For example, circuits 905 and 908 may be shift registers. Specifically, two clocks are provided to the LF PISO, and eight clocks with a phase difference of approximately 1U are provided to the HF PISO. Clock multiplexer 911 may delay the clocks to provide alignment between the HF PISO and driver 951. As its name indicates, the operating frequency of the LF PISO is lower than the operating frequency of the HF PISO. The driver includes digital-to-analog converters in multiple sub-blocks or slices.

[0062] Phase generator 924 generates two groups of 4-UI phases for a total of eight clocks. Each group includes four 1-UI spaced clocks. There are four positive phase (PPH) clocks on path 925 with a duty cycle of ~25% (1UI), and there are four negative phase (NPH) clocks on path 926 with a duty cycle of ~75% (3UI). A clock has a positive phase or a negative phase relative to another clock when it leads or lags, respectively. See also Fig. 10A .

[0063] Clock buffer 927 includes a set of buffers for distributing the PPH and NPH clocks from the phase generator to driver 951 via paths 928 and 929, respectively. The driver receives a data signal from the data path and a clock signal from the clock path at input node 961. Phase sampler 940 is coupled to paths 928 and 929 to compare the phases of the clocks and provide a corresponding output to digital control circuit 936 on path 937. In response, the digital control circuit provides a control signal to phase generator 924 on path 937a to adjust the phase. The digital control circuit also provides a control signal (e.g., SW1 and SW2) to switches SW1 and SW2 in the multiplexer segment of the phase sampler on path 941. Fig. 11B See also sel0 in Fig. 11B .

[0064] A clock-to-data phase detector 950, an average duty cycle sensor 952, and a target duty cycle sensor 955 are provided within the driver 951. The phase detector 950 detects the phase difference between the data signal on path 913 and the clock signals on paths 928 and 929, and provides a corresponding output to the digital control circuit 936 on path 939. In response, the digital control circuit provides a control signal to adjust the clock multiplexer 911 on path 939a.

[0065] The average duty cycle sensor 952 and the target duty cycle sensor each provide sensed duty cycle values ​​and target duty cycle values ​​for the PPH and NPH clocks (individually). Fig.13A and Fig. 13B During PPH / NPH calibration, the phases within the group are evenly spaced at 1-UI spacing and the duty cycle is set to its target value. The average duty cycle sensor 952 provides a corresponding output (indicative of the error in the duty cycle) to the digital control circuit 936 on path 938. In response, the digital control circuit provides a control signal to adjust the phase generator 924 on path 938a.

[0066] The data path source includes a digital (DIG) circuit / module 903 that receives the parallel data streams on paths 901 and 902 and applies feed forward equalization (FFE). For example, a 7-bit digital-to-analog converter (DAC) and an 8-tap FFE may be used. The digital (DIG) circuit / module 903 may be a source of digital data. The example transmitter uses PAM4 encoding, where every two bits represent one of four encoding levels (e.g., 00, 01, 11, or 10). In an example embodiment, the DIG generates a 64x7 bit word for every 64 transmitted symbols on path 904. The 64x7 bit word is multiplexed into an 8x7 bit word at an 8:1 multiplexer 906 in the LF PISO and output on path 907. The HF PISO further multiplexes the 8x7 bit word into a 4x7 bit word at a 2:1 multiplexer 909 and outputs it to the driver 951 on path 913. The multiplexer-based PISO advantageously consumes less power than shift register or FF-based serialization.

[0067] The clock to data phase detector (placed inside the driver) measures the skew of the data signal relative to the PPH (or NPH) clock. The adjustment algorithm adjusts the delay of the clock multiplexer 911 to set the skew to around 1.5UI, for example. This eliminates the requirement to re-time the latch at the HF PISO output.

[0068] Driver 951 provides the final 4:1 multiplexing, converting the 4x7 bit word into four symbols, driving differentially via output node 962, and delivering paths 953 and 954 to the output pads through the output network. For example, the driver in the example is a 7-bit digital-to-analog converter. In general, the DAC can be of any size and this depends on the output word from the DIG module.

[0069] In summary, transmitter 900 has many advantageous features, including a bootstrapped switch based clock sensor multiplexer in the phase sampler 940 allowing small device size but with small random mismatch, driver 951 slicing for generating optimal duty cycle targets for the driver clock, and an analog clock to data phase detector 950 allowing sub-UI data alignment.

[0070] Advantages include using multiplexers in the phase sampler. This can provide power reduction by reducing the load presented by the sensor on the clock network and mitigate single event jitter introduced by sensor switching.

[0071] Another advantage is the optimal duty cycle targeting. This improves jitter and SNDR as well as output driver common mode noise reduction.

[0072] Another advantage is the clock-to-data phase detector, which has sub-UI alignment accuracy and larger data alignment margin. This reduces the risk of data misalignment due to PVT variations and reduces power consumption by reducing the supply voltage in the data path.

[0073] Fig. 10A Depicted are various embodiments of Fig. 9 Examples of consistent PPH and NPH clock signals. Examples of PPH clocks include clk_pph1, clk2_pph, clk3_pph, and clk4_pph, where clk2_pph leads clk1_pph by 1UI, clk3_pph leads clk2_pph by 1UI, and clk4_pph leads clk3_pph by 1UI. Examples of NPH clocks include clk_nph1, clk2_nph, clk3_nph, and clk4_nph, where clk2_nph lags clk1_nph by 1UI, clk3_nph lags clk2_nph by 1UI, and clk4_nph lags clk3_nph by 1UI. In some cases, clk_pph1 is the same as clk_nph1.

[0074] The clock period of both PPH and NPH is 4UI. In PPH, the positive pulse width is 1UI (duty cycle is 25%), but in NPH, the positive pulse width is 3UI (duty cycle is 75%). The skew between phases after calibration (e.g., rising edge to rising edge) is 1UI.

[0075] Fig. 10B Depicted are various embodiments, Fig. 9 951 outputs an example of a PAM4 signal. A PAM4 signal can have one of four levels, where each level represents two bits of data. This example includes bits 00, 10, 01, 10, 11, 10, and 01. PAM4 is an example of an analog multi-level signal that can be output from transmitter 900. The width of each symbol is 1UI.

[0076] Fig.11A1 Depicted are various embodiments, Fig. 9 An example block diagram of a phase sampler 940 is shown. The phase sampler provides the following types of readings. The first type of reading is the duty cycle for each phase. The second type of reading is the PPH to NPH alignment, such as the alignment of the PPH clock to the NPH clock. The third type of reading is the alignment within the PPH (e.g., PPHR2R), which measures the spacing between PPH rising edges. R2R means rising edge to rising edge. The fourth type of reading is the alignment within the NPH (e.g., NPH F2F), which measures the spacing between NPH falling edges. F2F means falling edge to falling edge.

[0077] The phase sampler includes: a set of multiplexers 1010-1013, which receive at their inputs the clock signals from paths 928 and 929 in the clock path 960 of the transmitter; and a set of phase sensors 1020-1022 coupled to the outputs of the set of multiplexers.

[0078] Input 4:1 multiplexers 1010, 1011, 1012, and 1013 can be used to select phase pairs (clocks with different phases). Specifically, both multiplexers 1010 and 1011 receive four PPH clock signals on path 928, and both multiplexers 1012 and 1013 receive four NPH clock signals on path 929. Each multiplexer selects one clock signal at a time and provides the clock signal to one of the sensors at its output. Each sensor receives two clock signals at a time and compares the phases of the two clock signals to determine the phase difference. Once the comparison for a pair of clock signals is completed, the next pair of clock signals can be selected until all combinations of clock signals have been compared. The sensor can transmit the phase difference to the digital control unit 936 for adjusting the timing of the phase generator 924.

[0079] The sensors include a PPH R2R sensor 1020, a P2N sensor 1021, and an NPH F2F sensor 1022. Specifically, the PPH R2R sensor receives the PPH clock outputs of multiplexers 1010 and 1011, and determines the phase difference between the PPH clock outputs based on the time difference between the rising edges of the PPH clock outputs. The P2N sensor receives the PPH and NPH outputs of multiplexers 1011 and 1013, respectively. P2N means positive to negative. The NPH F2F sensor receives the NPH outputs of multiplexers 1012 and 1013, and determines the phase difference between the NPH outputs based on the time difference between the falling edges of the NPH outputs. The duty cycle sensor 1023 also receives the PPH and NPH clocks. The duty cycle sensor 1023 is just an RC filter, while the average duty cycle sensor 952 includes active devices, such as Fig. 13B shown.

[0080] The outputs of sensors 1020-1023 are multiplexed in an 11:1 output multiplexer. The 11:1 multiplexer 1030 is used to output the DC analog signal from the phase sampler 940 to the digital control circuit 936 for adjusting the phase generator 924. The DC analog signal is provided to a 1-bit ADC (comparator and Vref), the output of which is provided to the digital control section.

[0081] Next, an example implementation of one of the multiplexers 1010 to 1013 is discussed. The multiplexers may have the same configuration.

[0082] Fig.11A2 Depicted are various embodiments of Fig.11A1 An example timing diagram associated with a P2N sensor 1021 is shown. The timing diagram depicts the inputs PPH and NPH and the output P2N of the P2N sensor. When PPH transitions high, P2N transitions high; and when NPH transitions low, P2N transitions low. Time t represents the time between these transitions. The P2N sensor can be used when the time difference between the phases is on the order of 2UI. When the PPH phase differs from the NPH phase by exactly 2UI, the output of the gate is a clock signal with a duty cycle of 50%.

[0083] Fig. 11B Depicted are various embodiments, Fig.11A1 An example implementation of a multiplexer 1013 of FIG. The multiplexer includes four sections 1110, 1120, 1130, and 1140, which receive clock signals ph0, ph1, ph2, and ph3, respectively, and output signals Y0, Y1, Y2, and Y3, respectively. When the section front outputs Y0, Y1, Y2, and Y3 are shorted to a single line, a single output Y is provided. This provides a 4-to-1 multiplexer with 4 clock inputs, 4 select controls, and a single output Y. For multiplexers 1010 and 1011, ph0, ph1, ph2, and ph3 can represent PPH clock signals. For multiplexers 1012 and 1013, ph0, ph1, ph2, and ph3 can represent NPH clock signals. Section 1110 is described in detail.

[0084] The 4:1 multiplexer 1013 is based on a bootstrapped nMOS switch (transistor M1). Each multiplexer section includes switches (SW1 and SW2), a capacitor C, and a resistor R. The clock signal ph0 is input on a path 1101, which is coupled to the first switch SW1 and the source (S) of the nMOS transistor M1. Path 1101 is the input path of one of the clock signals. The drain (D) of the nMOS transistor is coupled on an output path 1102 to provide an output signal Y.

[0085] SW1 is connected in series with capacitor C in path 1103 and is controlled by a first select signal sel0, for example, from digital control circuit 936. Path 1103 is connected in parallel with a control gate path 1105 coupled to the control gate (G) of M1. These two paths are in turn coupled to path 1104, which includes a resistor R. Path 1104 is coupled to a second switch SW2, which is also controlled by sel0. In a first position, SW2 connects path 1104 to a high voltage at node 1106 and power supply node 1108. In a second position, SW2 connects path 1104 to a ground voltage at node 1107 and ground node 1109.

[0086] The nMOS transistor M1 is bootstrapped because the voltage on its control gate can increase the voltage stored on the capacitor. When M1 is turned on (on), path 1101 is coupled to path 1102, in which case Y0=ph0, for example, a high voltage or a low (0V) voltage. When SW2 couples paths 1104 and 1105 to the power supply node, M1 is turned on. When SW2 couples paths 1104 and 1105 to ground, M1 is turned off (not conducting). In this case, path 1102 has a floating voltage and a high impedance (Z).

[0087] When SW1 is closed (conducting), the bottom side of the capacitor is coupled to path 1101. When SW1 is open (not conducting), the bottom side of the capacitor has a floating voltage.

[0088] Sections 1120 , 1130 , and 1140 may have similar configurations as section 1110 , except that their switches will be controlled by control signals sel1 , sel2 , and sel3 , respectively.

[0089] FIG. 12A to FIG. 12D Example timing diagrams are shown for multiplexer sections 1110, 1120, 1130, and 1140, respectively. As previously mentioned, the multiplexer has a single output Y.

[0090] Fig. 12A Depicted are various embodiments of Fig. 11B 1 and 2. Figure 1 is an example timing diagram consistent with the multiplexer portion 1110 of FIG. 1. When sel0 is high, the portion is active. ph0 goes high at t0 and t1 for a duration of Δt(1). sel0 goes low at t2. During the entire time that portion 1110 is selected, sel1-3 are low (0V), indicating that the associated portion is inactive. Vg_M1, the control gate voltage of M1, and also the voltage on path 1105, follows ph0, going high at t0 and t1 for a duration of Δt. When sel0 transitions lower, Vg_M1 drops slightly at t2. Y follows ph0, going high at t0 and t1 for a duration of Δt, and then transitions to the high Z state when sel0 transitions low and the portion becomes inactive.

[0091] When sel0=1, SW1 is closed and SW2 is connected to the high voltage supply (HV) in the multiplexer section 1110. The HV level should be chosen so that Vgs(M1) is as high as possible (i.e., about 1V in this case), but small enough to avoid electrical overstress (EOS) on transistor M1. Note that the HV value depends on the duty cycle and the level of ph0, so different HV levels will be used for NPH and PPH sensors. MOS devices with high Vgs are less sensitive to Vth variations (the main source of device mismatch). A "conventional" (i.e., complementary metal oxide semiconductor (CMOS) or tri-state) multiplexer will have its devices switching between the on and off states during the rise / fall of ph0, which means that most of the time, the devices have a lower Vgs (and are therefore sensitive to Vth variations).

[0092] When sel0=0, Vg(M1)=0, so the output is at high impedance. Multiplexers may experience small kickbacks during switching due to relatively small device size and long transition times of gate voltage. Vg_M1 drops slightly at t2 because there is a path from Vg_M1 to ground when sel0=0 SW2 is grounded. When sel0=1, Vgs of M1 is almost constant and is determined by the voltage difference of the capacitors (close to 1V).

[0093] Fig. 12B Depicted are various embodiments of Fig. 11B 1+Δt and t1+Δt for duration Δt. sel1 goes low at t2+Δt. During the entire time that section 1120 is selected, sel0, sel2, sel3 are low at 0V. Vg_M1 follows ph1 going high at t0+Δt and t1+Δt for duration Δt. When sel1 transitions lower, Vg_M1 drops slightly at t2+Δt. Y follows ph1 going high at t0+Δt and t1+Δt for duration Δt, then transitions to the high-Z state when sel1 transitions low. When sel1=1, SW1 is closed and SW2 is connected to the high voltage supply (HV) in the multiplexer section 1120.

[0094] Fig. 12C Depicted are various embodiments of Fig. 11BExample timing diagram consistent with the multiplexer portion 1130 of FIG. When sel2 is high, the portion is active. ph2 goes high at t0+2Δt and t1+2Δt for a duration of Δt. sel2 goes low at t2+2Δt. During the entire time of the selection portion 1130, sel0, sel1, sel3 are low at 0V. Vg_M1 follows ph2 and goes high at t0+2Δt and t1+2Δt for a duration of Δt. When sel2 transitions to lower, Vg_M1 drops slightly at t2+2Δt. Y follows ph2, goes high at t0+2Δt and t1+2Δt for a duration of 2Δt, and then transitions to the high-Z state when sel2 transitions to low. When sel2=1, SW1 is closed and SW2 is connected to the high voltage power supply (HV) in the multiplexer portion 1130.

[0095] Fig.12D Depicted are various embodiments of Fig. 11B Example timing diagram consistent with the multiplexer portion 1140 of FIG. When sel3 is high, the portion is active. ph3 goes high at t0+3Δt and t1+3Δt for a duration of Δt. sel3 goes low at t2+3Δt. During the entire time of the selection portion 1140, sel0, sel1, sel2 are at a low level, at 0V. Vg_M1 follows ph3 and goes high at t0+3Δt and t1+3Δt for a duration of Δt. When sel3 transitions to low, Vg_M1 drops slightly at t2+3Δt. Y follows ph3 and goes high at t0+3Δt and t1+3Δt for a duration of 3Δt, and then transitions to the high-Z state when sel3 transitions to low. When sel3=1, SW1 is closed and SW2 is connected to the high voltage power supply (HV) in the multiplexer portion 1140.

[0096] Fig.13A Depicted are various embodiments, Fig. 9 1. Example implementation of a target duty cycle sensor 955. Sensor 955 includes a NAND gate 1310 that receives <0,0,0,1> on a first input path 1311 and receives d_dc on a second input path 1312. NOR gate 1320 receives <1,1,1,0> on a first input path 1321 and receives d_dc on a second input path 1322. <0,0,0,1> and <1,1,1,0> are logic inputs of a 4-bit bus. The four instances of the block / sensor 955 shown can be connected in parallel. d_dc is set in the digital control section 936.

[0097] The output of the NAND gate is provided to the control gate of the pMOS transistor 1331, and the output of the NOR gate is provided to the control gate of the nMOS transistor 1333. The pMOS and nMOS transistors are coupled in a series path that includes a power supply node 1330, for example 0.9V, a midpoint 1332 between the transistors and ground. Path 1341 is coupled to the midpoint 1332 via resistors Rslc and Rs and carries the target value drv_dc_tgt. The path is also coupled to an inverter 1340 that receives d_dc as an input.

[0098] exist Fig.13A In the example, the drv_dc_tgt value of NPH is obtained when d_dc = "1", and the drv_dc_tgt value of PPH is obtained when d_dc = "0". When the transmitter is operating, the optimal incremental pulse width is always tracked because it may be affected by temperature and power supply changes during operation. Not only this sensor is used to track temperature and power supply changes, but all sensors in the phase sampler block 940 and the phase detector 950 are also used in this way. For example, temperature changes can vary on a time scale of milliseconds or microseconds.

[0099] Fig. 13B Depicted are various embodiments, Fig. 9 An example implementation of an average duty cycle sensor 952. The sensor has Fig.13A The sensor in the same structure, but uses different signals. Fig.13A The difference is that NAND gate 1310 receives PPH<3:0> on first input path 1311, NOR gate 1320 receives NPH<3:0> on first input path 1321, and path 1341 carries the actual average duty cycle drv_dc_sns. A low pass filter (RC circuit) is attached to drv_dc_sns for generating its average signal.

[0100] Typically, the NPH / PPH duty cycle defines the active time of a driver slice. When the pulse width of PPH (NPH) is too short, there may be a time interval in which all four slices are disabled. In this case, the effective output impedance increases. On the other hand, when the pulse width is too long, more than two slices may be activated at the same time, which reduces the effective output impedance. Therefore, there is a certain pulse width during which the output impedance exactly matches the output impedance of the continuously enabled single slice.

[0101] In one approach, the duty cycle sensor can determine the duty cycle of one clock signal at a time. The sensor can communicate the result to the digital control 936 for use in adjusting the timing of the phase generator 924.

[0102] exist Fig.13A and Fig. 13B In the DAC, d_dc is the voltage that defines whether the high side or low side is measured. The high side refers to measuring PPH<3:0>, while the low side refers to measuring NPH<3:0>. Fig.13A The circuit has a continuously enabled slice, generating the target value drv_dc_tgt, while Fig. 13B The circuit is driven by NPH / PPH and measures the actual average duty cycle drv_dc_sns. The average duty cycle can be increased or decreased based on the comparison of the two values. To achieve the increase or decrease, the average duty cycle sensor 952 communicates with the digital control part 936, which in turn communicates with the Fig. 9 The phase generator 924 communicates with each other.

[0103] SLC 8LSB refers to the 8 least significant bits (LSB) strength of the replica compared to the 128 LSBs of the entire drive (7-bit DAC).

[0104] Notice, Fig.13A and Fig. 13B Two separate circuits are provided having the same structure but different inputs and outputs.

[0105] Fig.14A Depicted are various embodiments, Fig. 13B An example graph of the target and sensed output voltage of the positive phase (PPH) average duty cycle sensor 952 relative to the incremental pulse width (Δw).

[0106] Fig. 14B Depicted are various embodiments, Fig. 13B An example graph of the target and sensed output voltage of the negative phase (NPH) average duty cycle sensor 952 relative to the incremental pulse width (Δw).

[0107] These figures show the output voltage of the duty cycle sensor for a range of incremental pulse widths between 1U-1ps and 1UI+1ps, for example, 1UI=17.2ps at a 116Gbps data rate transmission. An optimal or otherwise preferred or desired pulse width can be determined at the point 1400 or 1410 where the two lines intersect, where the constant solid line represents drv_dc_tgt and the varying, dashed line represents drv_dc_sns. In one approach, the duty cycle sensor determines the optimal pulse width increment as the pulse width increment at which the measured duty cycle corresponds to the target duty cycle.

[0108] exist Fig.14A In the case of PPH duty cycle calibration, the optimal incremental pulse width is UI+0.75pSec, as shown by the dotted arrow, where Δw_opt is determined to be +0.75pSec.

[0109] exist Fig. 14B In the case of NPH duty cycle calibration, the optimal pulse width is 3UI-0.25pSec, as shown by the dotted arrow, where Δw_opt is determined to be -0.25pSec.

[0110] In either case, the "narrow" part of the pulse becomes wider.

[0111] Fig.15 Depicted are various embodiments, Fig. 9 The phase detector includes four paths from HF PISO ( Fig. 9 ) receives four data signals on path 913 and outputs the data to 4:1 multiplexer 1550. The first path 1510 includes a power supply node 1511 and a control gate voltage NPH <0> , D0 and D2 to output the bit Y0 on the path 1515 to the multiplexer 1550. The second path 1520 includes a power supply node 1521 and a pMOS transistor 1512, 1513 and 1514 having control gate voltages NPH <1> , D0 and D2 to output the bit Y1 on the path 1525 to the multiplexer. The third path 1530 includes a power supply node 1531 and a pMOS transistor 1522, 1523 and 1524 having a control gate voltage NPH <2> , D0 and D2 to output the bit Y2 on the path 1535 to the multiplexer. The fourth path 1540 includes a power supply node 1541 and a pMOS transistor 1532, 1533 and 1534 having a control gate voltage NPH <3> , D0 and D2 to output bit Y3 on path 1545 to the multiplexer. Each path is an open drain NOR gate.

[0112] As mentioned above, the driver phase (NPH / PPH) is not relevant to the 8-UI drive HF PISO 908. Therefore, in order to provide correct data alignment, a phase detector can be used to read the skew between the data and the NPH / PPH clock within the driver. The phase detector includes four open-drain NOR gates (NPH[3:0]+D0+D2) connected to a resistive load (resistor 1551) through a 4:1 multiplexer 1550. The output voltage is averaged by an RC filter including resistor 1552 and capacitor 1553 to provide an output PD on path 1554. , where i = 0, 1, 2, 3. <0> , PD <1> , PD <2> and PD <3> 1510 , 1520 , 1530 , and 1540 , respectively. Note that D0 and D2 are two of the four bits received by phase detector 950 on path 913 .

[0113] Fig.16 Depicted are various embodiments of Fig.15 Example timing diagrams consistent with the phase detector 950 of FIG. 1600, 1610, 1620, and 1630 depict NPH clock signals N0, N1, N2, and N3, respectively. N0 transitions low at t3-t4, t7-t8, and t11-t12. N1 transitions low at t0-t1, t4-t5, and t8-t9. N2 transitions low at t1-t1, t5-t6, and t9-t10. N3 transitions low at t2-t3, t6-t7, and t10-t11. Waveforms 1640 and 1650 depict D2 and D0, respectively. Waveforms 1660, 1670, 1680, and 1690 depict Y0, Y1, Y2, and Y3, respectively.

[0114] The HF PISO gets the sequence (01011010b) from the DIG 903 block. These are the input logic bits provided to the input of the HFPISO from the digital controller (936). Bits D0 and D2 are the 8-UI clock pattern derived from the HF PISO on path 913, while N0-N3 are the NPH set (i.e., N0=NPH <0> 、N1=NPH <1> 、N2=NPH <2> And N3 = NPH <3> ). When the transitions of D[0,2] are aligned with the rising edges of N1 and N3, respectively, the average outputs Y0 and Y2 are equal (PD0=PD2). Note that the matching of the transitions of D0 and D2 with the rising edges of N1 and N3 means that the flat or stable periods of D0 and D2 match those of N0 and N2, i.e., the data is correctly aligned. For example, the arrow extending from the low transition of N0 to D0 appears at the moment of the midpoint of the flat period of D0 (in this case, the difference of 1.5UI between the falling edge of D0 to the falling edge of N0 and 1.5UI from the rising edge of N0 to the rising edge of D0), for example, between t3 and t4, where the flat period is t2-t5, and between t7 and t8, where the flat period is t6-t7. Furthermore, the arrows extending from the low transition of N2 to D2 appear at moments in the midpoint of the flat period of D2, e.g., between t1 and t2, where the flat period is between t0-t3, between t5 and t6, where the flat period is between t4-t7, and between t7 and t8, where the flat period is between t6-t9.

[0115] Consider the case where the data is not aligned. For example, assume that D0 and D2 transitions are shifted to the left as shown by dashed lines 1641, 1643, 1651, and 1653, respectively, compared to solid lines 1642, 1644, 1652, and 1654. Then, Y0 will be longer as shown by dashed line 1661 compared to solid line 1662, and Y2 will become shorter as shown by dashed line 1681 compared to solid line 1682. As a result, PD0>PD2. Symmetrically, shifting D0 and D2 to the right will result in PD0 <PD2。

[0116] Forcing PD0=PD2 is not sufficient, as this may also occur if D0 and D2 match N2 and N0. To eliminate this situation, it is also necessary to ensure that PD1>PD3. Supporting both conditions (PD1>PD3 and PD2==PD0) results in good alignment, i.e. NPH and PPH sampled data with setup and hold times of about 1.5UI between the HF PISO and the driver.

[0117] A similar diagram can be provided for a PPH clock signal, but with Fig.15 The nMOS devices in the NAND are connected to ground instead of power so that the effective logic gates in the NAND are between the PPH clock signal and D0 and D2, rather than between the NPH clock signal and D0 and D2 in the NOR.

[0118] Fig.17 Examples of components that may be present in computing system 1750 for implementing the techniques (eg, operations, processes, methods, and methodologies) described herein are shown.

[0119] Fig. 9 The transmitter 900 can communicate with one or more other components in the computing system or outside the computing system to provide data signals. Communications outside the computing system can be passed to the external device 1772 through the interface circuit module 1770.

[0120] A power supply 1700 , such as a voltage converter, may provide power to one or more components of a computing system 1750 .

[0121] The memory circuit module 1754 may store instructions and the processor circuit module 1752 may execute the instructions to perform the functions described herein.

[0122] The computing system 1750 may include any combination of hardware or logic components referenced herein. These components may be implemented as ICs, portions thereof, discrete electronic devices, or other modules, instruction sets, programmable logic or algorithms, hardware, hardware accelerators, software, firmware, or combinations thereof suitable for the computing system 1750, or as components otherwise incorporated into a chassis of a larger system. For one embodiment, at least one processor 1752 may be packaged with computing logic 1782 and configured to practice aspects of the various example embodiments described herein to form a packaged system (SiP) or a system on a chip (SoC).

[0123] System 1750 includes processor circuits in the form of one or more processors 1752. Processor circuit module 1752 includes circuit modules such as, but not limited to, one or more processor cores and one or more of the following: cache memory, low dropout voltage regulator (LDO), interrupt controller, serial interface such as SPI, I2C or general programmable serial interface circuit, real-time clock (RTC), timer counter (including interval timer and watchdog timer), general I / O, memory card controller (e.g., secure digital / multimedia card (SD / MMC)), interface, mobile industry processor interface (MIPI) interface and joint test access group (JTAG) test access port. In some embodiments, processor circuit module 1752 may include one or more hardware accelerators (e.g., the same or similar to acceleration circuit module 1764), which may be a microprocessor, a programmable processing device (e.g., FPGA, ASIC, etc.), etc. One or more accelerators may include, for example, computer vision and / or deep learning accelerators. In some embodiments, the processor circuit module 1752 may include an on-chip memory circuit module, which may include any suitable volatile and / or non-volatile memory, such as DRAM, SRAM, EPROM, EEPROM, flash memory, solid-state memory, and / or any other type of memory device technology, such as those discussed herein.

[0124] The processor circuit module 1752 may include, for example, one or more processor cores (CPUs), application processors, GPUs, RISC processors, Acorn RISC Machine (ARM) processors, CISC processors, one or more DSPs, one or more FPGAs, one or more PLDs, one or more ASICs, one or more baseband processors, one or more radio frequency integrated circuits (RFICs), one or more microprocessors or controllers, multi-core processors, multi-threaded processors, ultra-low voltage processors, embedded processors, or any other known processing elements, or any suitable combination thereof. The processor (or core) 1752 may be coupled to a memory / storage device or may include a memory / storage device, and may be configured to execute instructions stored in the memory / storage device to enable various applications or operating systems to operate on the platform 1750. The processor (or core) 1752 is configured to operate application software to provide specific services to users of the platform 1750. In some embodiments, the processor 1752 may be a dedicated processor / controller configured (or configurable) to operate according to various embodiments herein.

[0125] As an example, the processor 1752 may include a ArchitectureCore TM processors, such as i3, i5, i7, i9-based processors; Microcontroller processor, such as Quark TM 、Atom TM Or other MCU-based processors; processor, Processors may be obtained from Santa Clara, California However, any number of other processors may be used, such as one or more of the following: Advanced Micro Devices (AMD) Architecture, e.g. or Processor, Accelerated Processing Unit (APU), MxGPU, Processor, etc. The company's A5-A12 and / or S1-S4 processors, Snapdragon Technologies TM or Centriq TM Processor, Texas Instruments, Open MultimediaApplications Platform (OMAP) TM Processors; MIPS-based designs from MIPS Technologies, such as the MIPS Warrior M-class, Warrior I-class, and Warrior P-class processors; ARM-based designs licensed from ARM Holdings plc, such as the ARM Cortex-A, Cortex-R, and Cortex-M series processors; Cavium TM Provided by the company In some embodiments, the processor 1752 may be part of a system on chip (SoC), a system in package (SiP), a multi-chip package (MCP), etc., wherein the processor 1752 and other components are formed as a single integrated circuit or a single package, such as from Edison TM or Galileo TM SoC Board. Other examples of processor 1752 are mentioned elsewhere in this disclosure.

[0126] The system 1750 may include or be coupled to an acceleration circuit module 1764, which may be implemented by one or more AI / ML accelerators, neural computing sticks, neuromorphic hardware, FPGAs, arrangements of GPUs, one or more SoCs (including programmable SoCs), one or more CPUs, one or more digital signal processors, dedicated ASICs (including programmable ASICs), PLDs such as complex PLDs (CPLDs) or high-complexity PLDs (HCPLDs), and / or other forms of dedicated processors or circuit modules designed to perform one or more specialized tasks. These tasks may include AI / ML processing (e.g., including training, reasoning, and classification operations), visual data processing, network data processing, object detection, rule analysis, etc. In an FPGA-based implementation, the acceleration circuit module 1764 may include logic blocks or logic structures and other interconnected resources that may be programmed (configured) to perform various functions, such as the processes, methods, functions, etc. of the various embodiments discussed herein. In such an embodiment, the acceleration circuit module 1764 may also include a memory unit (e.g., EPROM, EEPROM, flash memory, static memory (e.g., SRAM, antifuse, etc.)) for storing logic blocks, logic structures, data, etc. in a LUT, etc.

[0127] In some embodiments, the processor circuit module 1752 and / or the acceleration circuit module 1764 may include hardware elements specifically customized for machine learning and / or artificial intelligence (AI) functions. In these embodiments, the processor circuit module 1752 and / or the acceleration circuit module 1764 may be or may include an AI engine chip, which once loaded with appropriate weights and training code can run many different kinds of AI instruction sets. Additionally or alternatively, the processor circuit module 1752 and / or the acceleration circuit module 1764 may be or may include an AI accelerator, which may be one or more of the aforementioned hardware accelerators designed for hardware acceleration of AI applications. For example, these processors or accelerators may be artificial intelligence (AI) GPU clusters, The company's developed tensor processing unit (TPU), Real AI Processor (RAP TM ), Nervana provided by the company TM Neural Network Processor (NNP), Movidius TM Myriad TM X Vision Processing Unit (VPU), based on PX TM GPU, General Provided NM500 chip, Hardware provided by the company 3. Provided by Epiphany TM In some embodiments, the processor circuit module 1752 and / or the acceleration circuit module 1764 and / or the hardware accelerator circuit module may be implemented as an AI acceleration coprocessor, for example Hexagon 685DSP, Imagination Technologies The PowerVR 2NX Neural Network Accelerator (NNA) provided by The Neural Engine core in the A11 or A12 Bionic SoC, The neural processing unit (NPU) in the HiSilicon Kirin 970 provided, etc. In some hardware-based implementations, the various subsystems of the system 1750 can be operated by corresponding AI acceleration coprocessors, AI GPUs, TPUs or hardware accelerators (e.g., FPGAs, ASICs, DSPs, SoCs, etc.), etc., which are configured with appropriate logic blocks, bitstreams, etc. to perform their respective functions.

[0128] System 1750 also includes system memory 1754. Any number of memory devices may be used to provide a given amount of system memory. By way of example, memory 1754 may be or include volatile memory such as random access memory (RAM), static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), Dynamic Random Access Memory and / or any other desired type of volatile memory device. Additionally or alternatively, memory 1754 may be or include non-volatile memory, such as read-only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable (EEPROM), flash memory, non-volatile RAM, ferroelectric RAM, phase change memory (PCM), flash memory, and / or any other desired type of non-volatile memory device. Access to memory 1754 is controlled by a memory controller. Individual memory devices may be any number of different package types, such as a single die package (SDP), a dual die package (DDP), or a quad die package (Q17P). Any number of other memory implementations may be used, such as different kinds of dual in-line memory modules (DIMMs), including but not limited to microDIMMs or MiniDIMMs.

[0129] The storage circuit module 1758 provides persistent storage of information such as data, applications, operating systems, etc. In an example, the storage device 1758 can be implemented via a solid state disk drive (SSDD) and / or a high-speed electrically erasable memory (commonly referred to as "flash memory"). Other devices that can be used for the storage device 1758 include flash memory cards, such as SD cards, microSD cards, XD picture cards, etc., and USB flash drives. In an example, the memory device may be or may include a memory device using chalcogenide glass, a multi-threshold level NAND flash memory, a NOR flash memory, a single-level or multi-level phase change memory (PCM), a resistive memory, a nanowire memory, a ferroelectric transistor random access memory (FeTRAM), an antiferroelectric memory, a magnetoresistive random access memory (MRAM) using memristor technology, a phase change RAM (PRAM), a resistive memory including metal oxide-based, oxygen vacancy-based, and conductive bridge random access memory (CB-RAM), or a spin transfer torque (STT)-MRAM, a device based on spin electronic magnetic junction memory, a device based on a magnetic tunnel junction (MTJ), a device based on a domain wall (DW) and a spin orbit transfer (SOT), a thyristor-based memory device, a hard disk drive (HDD), a micro HDD, a combination thereof, and / or any other memory. The memory circuit module 1754 and / or the storage circuit module 1758 may also incorporate a memory device from and Three-dimensional (3D) cross-point (XPOINT) memory.

[0130] The memory circuit module 1754 and / or the storage circuit module 1758 are configured to store computing logic 1783 in the form of software, firmware, microcode or hardware level instructions to implement the techniques described herein. The computing logic 1783 can be used to store working copies and / or permanent copies of programming instructions, or data used to create programming instructions, for various components of the system 1750 (e.g., drivers, libraries, application programming interfaces (APIs), etc.), the operating system of the system 1750, the operation of one or more applications, and / or for executing the embodiments discussed herein. The computing logic 1783 can be stored or loaded into the memory circuit module 1754 as instructions 1782 or data used to create instructions 1782, and then the instructions 1782 or data used to create instructions 1782 are executed by the processor circuit module 1752 and accessed for execution to implement the functions described herein. The processor circuit module 1752 and / or the acceleration circuit module 1764 access the memory circuit module 1754 and / or the storage circuit module 1758 through the interconnect (IX) 1756. The instructions 1782 direct the processor circuit module 1752 to perform a specific sequence of actions or processes, for example, as described for the flowcharts and block diagrams of the operations and functions previously depicted. The various elements may be implemented by assembler instructions supported by the processor circuit module 1752 or a high-level language that can be compiled into instructions 1788 to be executed by the processor circuit module 1752 or data for creating instructions 1788. A permanent copy of the programming instructions may be placed in a permanent storage device of the storage circuit module 1758, for example, at the factory or in the field via a distributed medium (not shown), via a communication interface (e.g., from a distributed server (not shown)), wirelessly (OTA), or any combination thereof.

[0131] IX 1756 couples processor 1752 to communication circuit module 1766 for communicating with other devices such as a remote server (not shown). Communication circuit module 1766 is a hardware element or collection of hardware elements for communicating over one or more networks 1763 and / or with other devices. In one example, communication circuit module 1766 is or includes a transceiver circuit configured to be able to use any number of frequencies and protocols (e.g., Institute of Electrical and Electronics Engineers (IEEE) 802.11 (and / or its variants), IEEE 802.23.4, and / or Low power consumption (BLE), LoRaWAN TM The communication circuit module 1766 may be configured to communicate wirelessly using a cellular protocol such as 3GPP LTE and / or fifth generation (5G) / new radio (NR), etc. Additionally or alternatively, the communication circuit module 1766 may be or include one or more network interface controllers (NICs) for wired communication using, for example, an Ethernet connection, a controller area network (CAN), a local interconnect network (LIN), DeviceNet, ControlNet, Data Highway+, or PROFINET.

[0132] IX 1756 also couples processor 1752 to interface circuit module 1770, which is used to connect system 1750 to one or more external devices 1772. External devices 1772 may include, for example, sensors, actuators, positioning circuit modules (e.g., global navigation satellite system (GNSS) / global positioning system (GPS) circuit modules), client devices, servers, network devices (e.g., switches, hubs, routers, etc.), integrated photonic devices (e.g., optical neural network (ONN) integrated circuits (ICs), etc.), and / or other similar devices.

[0133] In some optional examples, various input / output (I / O) devices may be present within or connected to the system 1750, which is referred to as an input circuit module 1786 and an output circuit module 1784. The input circuit module 1786 and the output circuit module 1784 include one or more user interfaces designed to enable a user to interact with the platform 1750 and / or a peripheral component interface designed to enable a peripheral component to interact with the platform 1750. The input circuit module 1786 may include any physical or virtual unit for accepting input, including, in particular, one or more physical or virtual buttons (e.g., a reset button), a physical keyboard, a keypad, a mouse, a touchpad, a touch screen, a microphone, a scanner, a headset, and the like. The output circuit module 1784 may be included to display information or otherwise communicate information, such as sensor readings, actuator positions, or other similar information. Data and / or graphics may be displayed on one or more user interface components of the output circuit module 1784. Output circuit module 1784 may include any number and / or combination of audio or visual displays, including, in particular, one or more simple visual outputs / indicators (e.g., binary status indicators (e.g., light emitting diodes (LEDs)) and multi-character visual outputs, or more complex outputs, such as a display device or touch screen (e.g., a liquid crystal display (LCD), LED display, quantum dot display, projector, etc.), where the output of characters, graphics, multimedia objects, etc. is generated or produced by the operation of platform 1750. Output circuit module 1784 may also include speakers and / or other audio emitting devices, printers, etc. Additionally or alternatively, sensors may be used as Input circuit module 1786 (e.g., image capture device, motion capture device, etc.), and one or more actuators can be used as output device circuit module 1784 (e.g., actuators for providing tactile feedback, etc.). Peripheral component interfaces can include, but are not limited to, non-volatile memory ports, USB ports, audio jacks, power interfaces, etc. In some embodiments, in the context of the present system, display or console hardware can be used to provide output to the edge computing system and receive input to the edge computing system; manage components or services of the edge computing system; identify the status of edge computing components or services; or perform any other number of management or administrative functions or service use cases.

[0134] The components of the system 1750 may communicate via the IX 1756. The IX 1756 may include a variety of technologies including ISA, extended ISA, I2C, SPI, point-to-point interface, power management bus (PMBus), PCI, PCIe, PCIx, UPI, Accelerator Link, CXL, CAPI, OpenCAPI, QPI, UPI, OPA IX, RapidIO TM System IX, CCIX, Gen-Z Consortium IX, HyperTransport Interconnect, NVLink, Time Triggered Protocol (TTP) systems, FlexRay systems, PROFIBUS, and / or any number of other IX technologies provided. The IX1756 can be a dedicated bus, such as used in a SoC based system.

[0135] The number, capabilities, and / or capacity of the elements of system 1750 may vary, depending on whether computing system 1750 is used as a fixed computing device (e.g., a server computer in a data center, a workstation, a desktop computer, etc.) or a mobile computing device (e.g., a smartphone, a tablet computing device, a laptop computer, a game console, an IoT device, etc.). In various embodiments, computing device system 1750 may include one or more components of a data center, a desktop computer, a workstation, a laptop computer, a smartphone, a tablet computer, a digital camera, a smart appliance, a smart home hub, a network device, and / or any other device / system that processes data.

[0136] The techniques described herein may be performed in part or in whole by software or other instructions provided in a machine-readable storage medium (e.g., memory). The software is stored as processor-executable instructions (e.g., instructions for implementing any other process discussed herein). The instructions associated with the flowcharts (and / or various embodiments) and executed to implement embodiments of the disclosed subject matter may be implemented as part of an operating system or a specific application, component, program, object, module, routine, or other sequence of instructions or organization of sequences of instructions.

[0137] The storage medium may be a tangible, non-transitory machine-readable medium such as read-only memory (ROM), random access memory (RAM), flash memory devices, floppy disks and other removable disks, magnetic storage media, optical storage media (e.g., compact disk read-only memories (CD ROMS), digital versatile disks (DVDs)), etc.

[0138] The storage medium may be included in, for example, a communication device, a computing device, a network device, a personal digital assistant, a manufacturing tool, a mobile communication device, a cellular phone, a notebook computer, a tablet computer, a game console, a set-top box, an embedded system, a TV, or a personal desktop computer.

[0139] Some non-limiting examples of various embodiments are given below.

[0140] Example 1 includes an apparatus comprising: a data path; a clock path; a digital control section coupled to the data path and the clock path; and a driver coupled to the data path, the clock path, and the digital control section, wherein the driver includes a phase detector for detecting a phase error in a data signal from the data path, the driver is for transmitting the phase error to the digital control section, and the digital control section is for adjusting the timing of the data path based on the phase error.

[0141] Example 2 includes an apparatus according to Example 1, wherein, to adjust the timing of the data path based on the phase error, the digital control section is used to adjust the timing of a clock multiplexer that distributes a clock signal from the clock path to the data path.

[0142] Example 3 includes the apparatus of Example 2, wherein the data path comprises a digital data source, a low-frequency parallel-input serial-output (PISO) circuit coupled to the digital data source, and a high-frequency PISO circuit coupled to the low-frequency PISO circuit; and the clock multiplexer is to distribute the clock signal from the clock path to the low-frequency PISO circuit and the high-frequency PISO circuit.

[0143] Example 4 includes an apparatus according to any one of Examples 1-3, wherein: the driver includes a duty cycle sensor, which is used to: determine a target duty cycle and an average measured duty cycle, determine a pulse width increment based on the target duty cycle and the average measured duty cycle, and transmit the pulse width increment to the digital control unit; and the digital control unit is used to adjust the timing of the phase generator in the clock path based on the pulse width increment.

[0144] Example 5 includes the apparatus according to any one of Examples 1-4, further comprising an orthogonal sampler coupled to the output of the orthogonal generator in the clock path, wherein the orthogonal sampler is used to determine a phase error and a duty cycle error of a clock signal output from the orthogonal generator, and transmit the phase error and the duty cycle error to the digital control section, wherein the digital control section is used to adjust the timing of the orthogonal generator based on the phase error and the duty cycle error.

[0145] Example 6 includes an apparatus according to any one of Examples 1-5, further comprising a phase sampler in the clock path, wherein the phase sampler is used to determine a phase error and a duty cycle error of a clock signal in the clock path, and transmit the phase error and the duty cycle error to the digital control unit, and the digital control unit is used to adjust the timing of a phase generator in the clock path based on the phase error and the duty cycle error.

[0146] Example 7 includes an apparatus according to Example 6, wherein: the phase sampler includes a multiplexer, which receives the clock signal as an input; and the output of the multiplexer is coupled to a plurality of sensors, which are used to determine the phase error and duty cycle error of the clock signal.

[0147] Example 8 includes the apparatus of Example 7, wherein the plurality of sensors includes a sensor for determining the phase error of the clock signal based on at least one of: comparing a rising edge to a rising edge or comparing a falling edge to a falling edge.

[0148] Example 9 includes the apparatus of Example 7 or 8, wherein the plurality of sensors includes a sensor for determining the phase error of the clock signal based on a positive and negative comparison.

[0149] Example 10 includes an apparatus according to any of Examples 7-9, wherein: each of the multiplexers includes a collection of parts; each part includes a bootstrapped n-type metal oxide silicon field effect transistor (nMOSFET); the drain of the nMOSFET is coupled to an input path of one of the clock signals; and the source of the nMOSFET is coupled to an output path.

[0150] Example 11 includes an apparatus according to any of Examples 1-10, further comprising at least one of the following: a transmitter circuit, an integrated circuit, a system on a chip, a packaged system, or a computing device in which the data path, the clock path, the digital control unit, and the driver are provided.

[0151] Example 12 includes a device comprising: an input node for receiving a data signal from a data path and a clock signal from a clock path; a phase detector for detecting a phase error in the data signal and transmitting the phase error to a digital control unit; a duty cycle sensor for determining a target duty cycle and an average measured duty cycle, determining a pulse width increment based on the target duty cycle and the average measured duty cycle, and transmitting the pulse width increment to the digital control unit; and an output node for sending an analog multi-level signal based on the data signal and the clock signal.

[0152] Example 13 includes the apparatus of Example 12, wherein the digital control is to adjust timing of a phase generator in the clock path based on the pulse width increment.

[0153] Example 14 includes the apparatus of Example 12 or 13, wherein the digital control is to adjust timing of the data path based on the phase error.

[0154] Example 15 includes the apparatus of any of Examples 12-14, wherein the duty cycle sensor is configured to determine the pulse width increment as a pulse width increment where the measured duty cycle corresponds to a target duty cycle.

[0155] Example 16 includes an apparatus comprising: a set of multiplexers for receiving a clock signal from a clock path in a transmitter, wherein the clock signal is generated by a phase generator; a set of phase sensors coupled to the outputs of the set of multiplexers, wherein the phase sensors are used to determine a phase error of the clock signal; and an output multiplexer coupled to the outputs of the phase sensors, wherein the output multiplexer is used to provide the phase error to a digital control section, and the digital control section is used to adjust the timing of the phase generator based on the phase error.

[0156] Example 17 includes an apparatus according to Example 16, wherein: each multiplexer of the set of multiplexers comprises a collection of parts; and each part comprises a bootstrapped n-type metal oxide silicon field effect transistor (nMOSFET) coupled to an input path and an output path of one of the clock signals.

[0157] Example 18 includes an apparatus according to Example 17, wherein each portion includes a first switch and a second switch, the first switch being used to couple a path including a capacitor to the input path, and the second switch being used to couple a power supply node or ground to a control gate path of the nMOSFET and the path including the capacitor.

[0158] Example 19 includes an apparatus according to any one of Examples 16-18, wherein the clock signal includes a positive phase clock signal and a negative phase clock signal; and the set of phase sensors includes a sensor for determining a phase error of the positive phase clock signal based on a rising edge to a rising edge comparison and a sensor for determining a phase error of the negative phase clock signal based on a falling edge to a falling edge comparison.

[0159] Example 20 includes the apparatus of Example 19, wherein the set of phase sensors includes sensors for determining a positive and negative comparison of the negative phase clock signal.

[0160] Example 21 includes a method comprising: detecting a phase error in a data signal from a data path; transmitting the phase error to a digital control unit, wherein the digital control unit is coupled to the data path and the clock path; and at the digital control unit, adjusting the timing of the data path based on the phase error.

[0161] Example 22 includes the method of Example 21, wherein adjusting the timing of the data path based on the phase error comprises adjusting the timing of a clock multiplexer that distributes a clock signal from the clock path to the data path.

[0162] Example 23 includes the method of Example 22, further comprising: distributing the clock signal from the clock path to a low frequency PISO circuit and a high frequency PISO circuit in the data path.

[0163] Example 24 includes a method according to any of Examples 21-23, further comprising: determining a target duty cycle and an average measured duty cycle to determine a pulse width increment; transmitting the pulse width increment to the digital control unit; and adjusting the timing of the phase generator in the clock path based on the pulse width increment.

[0164] Example 25 includes the method according to any of Examples 21-24, further comprising: determining a phase error of a clock signal output from an orthogonal generator; transmitting the phase error to the digital control section; and at the digital control section, adjusting the timing of the orthogonal generator based on the phase error.

[0165] Example 26 includes a method according to any one of Examples 21-25, further comprising: determining a phase error of a clock signal in the clock path; transmitting the phase error to the digital control unit; and at the digital control unit, adjusting the timing of a phase generator in the clock path based on the phase error.

[0166] Example 27 includes the method according to Example 27 or 28, further comprising: determining the phase error of the clock signal based on a comparison of a falling edge to a falling edge.

[0167] Example 28 includes the method of Example 27 or 28, further comprising determining the phase error of the clock signal based on positive and negative comparisons.

[0168] Example 29 includes a non-transitory machine-readable storage including machine-readable instructions that, when executed, cause a processor or other circuit or computing device to implement the method according to any of Examples 21-28.

[0169] Example 30 includes a computer program comprising instructions that, when executed by a computer, cause the computer to perform the method according to any one of Examples 21 to 23.

[0170] Example 31 includes a method comprising: at a set of multiplexers, receiving a clock signal from a clock path in a transmitter, wherein the clock signal is generated by a phase generator; at a set of phase sensors coupled to the outputs of the set of multiplexers, determining a phase error of the clock signal; at an output multiplexer coupled to the outputs of the phase sensors, providing the phase error to a digital control unit; and at the digital control unit, adjusting the timing of the phase generator based on the phase error.

[0171] Example 32 includes the method of Example 31, further comprising coupling a path including a capacitor to the input path, and coupling a power supply node or ground to a control gate path of the nMOSFET and the path including the capacitor.

[0172] Example 33 includes a non-transitory machine-readable storage comprising machine-readable instructions that, when executed, cause a processor or other circuit or computing device to implement the method according to Example 31 or 32.

[0173] Example 34 includes a computer program comprising instructions that, when executed by a computer, cause the computer to perform the method according to Example 31 or 32.

[0174] In this detailed description, reference is made to the accompanying drawings, which form a part of the detailed description, in which the same reference numerals indicate the same parts throughout and are shown therein by way of illustrative embodiments that may be implemented. It should be understood that other embodiments may be used and structural or logical changes may be made without departing from the scope of the present disclosure. Therefore, the following detailed description should not be considered to have a limiting meaning, and the scope of the embodiments is defined by the appended claims and their equivalents.

[0175] Various operations can be described as a plurality of discrete actions or operations in sequence in a manner that is most helpful for understanding the claimed subject matter. However, the order of description should not be interpreted as implying that these operations must be dependent on the order. These operations may not be performed in the order presented. The described operations may be performed in an order different from the described embodiments. In other embodiments, various additional operations may be performed and / or the described operations may be omitted.

[0176] The terms "substantially", "close", "approximately", "close to", and "about" generally refer to within + / -10% of a target value. Unless otherwise specified, the use of ordinal adjectives "first", "second", and "third", etc. to describe common objects merely indicates that different instances of similar objects are referred to, and are not intended to imply that the objects so described must be in a given order, whether in time, space, in order, or in any other manner.

[0177] For purposes of this disclosure, the phrases "A and / or B" and "A or B" mean (A), (B), or (A and B). For purposes of this disclosure, the phrases "A, B, and / or C" mean (A), (B), (C), (A and B), (A and C), (B and C), or (A, B, and C).

[0178] This specification may use the phrases "in an embodiment" or "in some embodiments," which may each refer to one or more of the same or different embodiments. In addition, as used with respect to the embodiments of the present disclosure, the terms "comprising," "including," "having," and the like are synonymous.

[0179] As used herein, the term "circuitry" may refer to, be a part of, or include an application specific integrated circuit ("ASIC"), an electronic circuit, a processor (shared, dedicated, or group), a combinational logic circuit, and / or other suitable hardware components that provide the described functionality. As used herein, a "computer-implemented method" may refer to a method implemented by one or more processors, a computer system having one or more processors, a mobile device such as a smartphone (which may include one or more processors), a tablet computer, a laptop computer, a set-top box, a game console, etc.

[0180] The terms "coupled", "communicative coupling" and their derivatives are used herein. The term "coupled" may refer to two or more elements being in direct physical or electrical contact with each other, may refer to two or more elements being in indirect contact with each other but still cooperating or interacting with each other, and / or may refer to one or more other elements being coupled or connected between elements referred to as being coupled to each other. The term "directly coupled" may mean that two or more elements are in direct contact with each other. The term "communicative coupling" may refer to that two or more elements may be in contact with each other by means of communication including by wired or other interconnect connections, by wireless communication channels or links, and the like.

[0181] Reference in this specification to "an embodiment," "one embodiment," "some embodiments," or "other embodiments" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least some embodiments, but not necessarily in all embodiments. Various appearances of "an embodiment," "one embodiment," or "some embodiments" do not necessarily refer to the same embodiment. If the specification states that a component, feature, structure, or characteristic "may," "might," or "could" be included, that component, feature, structure, or characteristic is not required to be included. If the specification or claims refer to "a" or "an" element, this does not mean that there is only one element. If the specification or claims refer to "additional" elements, this does not preclude the presence of more than one additional element.

[0182] In addition, in one or more embodiments, features, structures, functions or characteristics can be combined in any suitable manner. For example, in any case where the specific features, structures, functions or characteristics associated with the two embodiments are not mutually exclusive, the first embodiment can be combined with the second embodiment.

[0183] Although the present disclosure has been described in conjunction with the specific embodiments of the present disclosure, many substitutions, modifications and variations of such embodiments will be apparent to those skilled in the art from the foregoing description. The embodiments of the present disclosure are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims.

[0184] In addition, for simplicity of illustration and discussion, and so as not to obscure the present disclosure, well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the presented figures. In addition, the arrangements may be shown in block diagram form to avoid the present disclosure, and also in view of the fact that the details of the implementation of these block diagram arrangements are highly dependent on the platform in which the present disclosure will be implemented (i.e., these details should be within the scope of those skilled in the art). Wherein, specific details (e.g., circuits) are set forth to describe example embodiments of the present disclosure, it should be apparent to those skilled in the art that the present disclosure can be implemented with or without variations of these specific details. This specification is therefore considered to be illustrative rather than restrictive.

[0185] The abstract provided will enable the reader to ascertain the nature and gist of the technical disclosure. The abstract is submitted with the understanding that it will not be used to limit the scope or meaning of the claims. Therefore, the following claims are hereby incorporated into the detailed description, with each claim acting as a separate embodiment by itself.

Claims

1. A device, include: Data path; Clock path; a digital control section coupled to the data path and the clock path; as well as A driver is coupled to the data path, the clock path and the digital control section, wherein the driver includes a phase detector for detecting a phase error in a data signal from the data path, the driver is for transmitting the phase error to the digital control section, and the digital control section is for adjusting the timing of the data path based on the phase error.

2. The device according to claim 1, in, To adjust the timing of the data path based on the phase error, the digital control section is configured to adjust the timing of a clock multiplexer that distributes a clock signal from the clock path to the data path.

3. The device according to claim 2, in: The data path includes a digital data source, a low frequency parallel-input serial-output (PISO) circuit coupled to the digital data source, and a high frequency PISO circuit coupled to the low frequency PISO circuit; and The clock multiplexer is used to distribute the clock signal from the clock path to the low-frequency PISO circuit and the high-frequency PISO circuit.

4. The device according to claim 1 or 2, in: The driver includes a duty cycle sensor for: determining a target duty cycle and an average measured duty cycle, determining a pulse width increment based on the target duty cycle and the average measured duty cycle, and transmitting the pulse width increment to the digital control portion; as well as The digital control section is used to adjust the timing of a phase generator in the clock path based on the pulse width increment.

5. The apparatus of claim 1 or 2, further comprising a quadrature sampler coupled to an output of a quadrature generator in the clock path, in, The quadrature sampler is used to determine a phase error and a duty cycle error of a clock signal output from the quadrature generator, and transmit the phase error and the duty cycle error to the digital control part, wherein the digital control part is used to adjust the timing of the quadrature generator based on the phase error and the duty cycle error.

6. The apparatus according to claim 1 or 2, further comprising a phase sampler in the clock path, in, The phase sampler is used to determine a phase error and a duty cycle error of a clock signal in the clock path, and transmit the phase error and the duty cycle error to the digital control section, and the digital control section is used to adjust the timing of a phase generator in the clock path based on the phase error and the duty cycle error.

7. The device according to claim 6, in: The phase sampler includes a multiplexer that receives the clock signal as an input; and The output of the multiplexer is coupled to a plurality of sensors for determining a phase error of the clock signal.

8. The device according to claim 7, in, The plurality of sensors includes a sensor for determining the phase error and the duty cycle error of the clock signal based on at least one of: a rising edge to a rising edge comparison or a falling edge to a falling edge comparison.

9. The device according to claim 7 or 8, in, The sensor comprises a sensor for determining the phase error of the clock signal based on a positive and negative comparison.

10. The device according to claim 7 or 8, in: Each of the multiplexers comprises a collection of parts; Each section includes a bootstrapped n-type metal oxide silicon field effect transistor (nMOSFET); The drain of the nMOSFET is coupled to an input path of one of the clock signals; and The source of the nMOSFET is coupled to an output path.

11. The apparatus according to claim 1 or 2, further comprising at least one of the following: a transmitter circuit, an integrated circuit, a system on a chip, a packaged system, or a computing device in which the data path, the clock path, the digital control section, and the driver are provided.

12. A device, include: An input node for receiving a data signal from a data path and a clock signal from a clock path; a phase detector for detecting a phase error in the data signal and transmitting the phase error to a digital control unit; a duty cycle sensor for: determining a target duty cycle and an average measured duty cycle, determining a pulse width increment based on the target duty cycle and the average measured duty cycle, and transmitting the pulse width increment to the digital control portion; as well as An output node is used to send an analog multi-level signal based on the data signal and the clock signal.

13. The device according to claim 12, in, The digital control is for adjusting the timing of a phase generator in the clock path based on at least one of: the pulse width increment or the phase error.

14. The device according to claim 12 or 13, in, The duty cycle sensor is used to determine the pulse width increment as a pulse width increment in which the measured duty cycle corresponds to the target duty cycle.

15. A device, include: a set of multiplexers for receiving a clock signal from a clock path in the transmitter, wherein the clock signal is generated by a phase generator; a set of phase sensors coupled to outputs of the set of multiplexers, wherein the phase sensors are used to determine a phase error of the clock signal; and An output multiplexer is coupled to the output of the phase sensor, wherein the output multiplexer is configured to provide the phase error to a digital control, and the digital control is configured to adjust the timing of the phase generator based on the phase error.

16. The device according to claim 15, in: Each multiplexer in the set of multiplexers comprises a collection of parts; and Each section includes a bootstrapped n-type metal oxide silicon field effect transistor (nMOSFET) coupled to an input path and an output path of one of the clock signals.

17. The device according to claim 16, in: Each section includes a first switch for coupling a path including a capacitor to the input path and a second switch for coupling a power supply node or ground to a control gate path of the nMOSFET and the path including the capacitor.

18. The device according to claim 15 or 16, in: The clock signal includes a positive phase clock signal and a negative phase clock signal; and The set of phase sensors includes a sensor for determining a phase error of the positive phase clock signal based on a rising edge to rising edge comparison and a sensor for determining a phase error of the negative phase clock signal based on a falling edge to falling edge comparison.

19. The device according to claim 18, in, The set of phase sensors includes a sensor for determining a positive and negative comparison of the negative phase clock signal.

20. The device according to claim 15 or 16, in, The clock signal includes a positive phase clock signal and a negative phase clock signal, and the apparatus further includes a set of duty cycle sensors for both the positive phase clock signal and the negative phase clock signal.

21. A method, include: detecting a phase error in a data signal from a data path; transmitting the phase error to a digital control, wherein the digital control is coupled to the data path and the clock path; as well as At the digital control section, the timing of the data path is adjusted based on the phase error.

22. A non-transitory machine-readable storage device comprising machine-readable instructions, which, when executed, cause a processor or other circuit or computing device to perform a method, the method include: detecting a phase error in a data signal from a data path; transmitting the phase error to a digital control, wherein the digital control is coupled to the data path and the clock path; as well as At the digital control section, the timing of the data path is adjusted based on the phase error.

23. The non-transitory machine-readable storage device of claim 22, in, Adjusting the timing of the data path based on the phase error includes adjusting the timing of a clock multiplexer that distributes a clock signal from the clock path to the data path.

24. The non-transitory machine-readable storage device of claim 23, in, The method also includes distributing the clock signal from the clock path to a low-frequency PISO circuit and a high-frequency PISO circuit in the data path.

25. The non-transitory machine-readable storage device according to claim 22 or 23, in, The method further comprises: determining a target duty cycle and an average measured duty cycle to determine a pulse width increment; transmitting the pulse width increment to the digital control; and The timing of a phase generator in the clock path is adjusted based on the pulse width increment.

Citation Information

Cited By

  • Sending channel circuit based on 1 / 4 rate structure SERDES

    CN121935200A