High-speed data communication and noise suppression method for digital isolator

Abstract

The application relates to the technical field of electric digital data processing, and discloses a high-speed data communication and noise suppression method of a digital isolator; the application proposes a cross-layer cooperative optimization scheme aiming at the problem that the existing digital isolator is easily affected by noise during high-speed communication, leading to data distortion; the method comprises the following steps: at the sending end, an adaptive pre-emphasis technology is used to compensate for the high-frequency attenuation of the isolation channel, and the to-be-transmitted data is subjected to channel coding; at the receiving end, a dynamic threshold comparator is used to sample the signal, soft decision data containing amplitude information is obtained, and Viterbi decoding is carried out based on the soft information to correct transmission errors; and the application further combines a hybrid automatic repeat request mechanism to adaptively adjust the coding rate according to the channel quality; through joint optimization of the physical layer and the link layer, the application significantly improves the data transmission rate and reliability of the digital isolator under a high common-mode transient noise environment, and effectively guarantees the integrity of the data.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a method for high-speed data communication and noise suppression using a digital isolator. Background Technology

[0002] Digital isolators, as key components capable of transmitting signals between different voltage domains without direct electrical connection, play an indispensable role in modern electronic systems. Their core applications cover multiple high-value fields such as industrial automation, new energy vehicles, medical electronics, and communication infrastructure.

[0003] In industrial automation, programmable logic controllers (PLCs) and servo drives typically need to transmit pulse-width modulated signals and fault feedback signals between the processing core (such as MCUs and DSPs) and high-voltage power devices (such as IGBTs and MOSFETs). Digital isolators are responsible not only for level conversion but, more importantly, for protecting low-voltage control circuits from surge voltages and common-mode transient interference reaching hundreds or even thousands of volts in motor drive circuits.

[0004] In new energy vehicles, the digital isolator (DISA) serves as the core communication hub between the battery management system (BMS) and the traction inverter. The BMS needs to monitor the voltage and temperature of hundreds of series-connected cells in the battery pack in real time and report the data to the vehicle controller via the isolation barrier. Simultaneously, the traction inverter receives torque commands from the vehicle controller to drive the motor. These scenarios demand extremely high communication reliability; any communication interruption or data error caused by noise could lead to loss of vehicle power or even a safety accident.

[0005] In the field of medical electronics, devices such as electrocardiographs (ECGs) and electroencephalograms (EEGs) require electrical isolation between the data acquisition front-end (which comes into contact with the human body) and the back-end signal processing and power supply sections to ensure patient safety. Digital isolators are used to transmit weak physiological signals and must achieve high-precision signal transmission without introducing additional noise to ensure the accuracy of diagnostic results.

[0006] Despite existing digital isolator technologies, such as those based on magnetic coupling (iCoupler) or capacitive isolation (… Capacitor-based products have been widely used in the market, but in pursuit of higher data transmission rates (such as above 100Mbps) and stronger noise suppression capabilities, they still face the following technical challenges:

[0007] The first technical problem is the attenuation of high-frequency signals and inter-symbol interference (ISI). As data transmission rates increase, the frequency components of the signal become higher. The channel model of the isolation transformer or capacitor itself has bandpass characteristics, resulting in significant insertion loss at high frequencies. This causes high-speed pulses to be distorted when passing through the isolator, leading to mutual interference between consecutive symbols and forming ISI. Severe ISI can cause the eye diagram at the receiver to close, directly leading to sampling and decision errors and limiting further increases in communication rates.

[0008] The second technical problem is insufficient common-mode transient immunity. In high-voltage applications such as motor drives and switching power supplies, there are huge voltage fluctuations on both sides of the isolator, with common-mode voltage transients (dv / dt) reaching tens or even hundreds of kV / µs. This severe common-mode noise can couple to the signal path through the parasitic parameters of the isolation capacitor, forming differential-mode noise. Traditional digital isolators typically rely on comparators with fixed thresholds for decision-making. When the noise spike amplitude caused by the common-mode transient exceeds the signal amplitude, it is very easy to cause false triggering, resulting in spurious data transitions.

[0009] The third technical problem is its weak ability to handle sudden, long errors. Traditional isolators typically use only simple Manchester coding or basic digital buffering, lacking an effective channel coding mechanism. When encountering persistent strong interference (such as ESD discharge or power fluctuations caused by load changes), multiple bits may be flipped consecutively. Existing simple verification mechanisms (such as parity checking) can detect errors but cannot correct them, only requesting retransmission. In high-noise environments, frequent retransmissions can severely reduce effective throughput and even cause the communication link to freeze.

[0010] The fourth problem is the poor adaptability of isolators to changing channel characteristics due to environmental variations. The transmission characteristics of isolators are not static. Temperature changes can alter permeability or dielectric constant, affecting coupling efficiency; manufacturing process variations between different batches can also lead to differences in channel gain. Most existing products use fixed transmit drive strength and fixed receive thresholds, failing to adaptively adjust to the current channel conditions. This often results in system design sacrificing nominal performance to meet worst-case reliability requirements.

[0011] Therefore, those skilled in the art urgently need a method to achieve stable data transmission using digital isolators under non-ideal conditions. Summary of the Invention

[0012] The core technical problem addressed by this invention is: how to achieve high speed (≥100Mbps) and high reliability (bit error rate <100Mbps) for digital isolators in environments with strong common-mode noise, non-ideal channel characteristics, and time-varying conditions. Data transmission.

[0013] Solving the aforementioned core technical problems can provide a key technical foundation for the application of high-voltage SiC / GaN power devices and the implementation of high-resolution medical imaging systems. If high-speed communication under high noise immunity cannot be achieved, the switching frequency and efficiency of high-voltage systems will be limited, which will lead to the real-time transmission of high-resolution images being constrained by data bottlenecks and unable to meet increasingly demanding practical application requirements.

[0014] To this end, this application designs a comprehensive digital communication method that integrates physical layer optimization and data link layer coding. This method aims to break away from the traditional digital isolator's simple role as a "transparent buffer" and upgrade it into a communication node with intelligent signal processing capabilities.

[0015] The core content of this invention is as follows:

[0016] At the physical layer, this invention abandons the traditional fixed threshold comparison and introduces dynamic threshold tracking technology to fit the drift of the signal baseline in real time, and uses adaptive pre-emphasis technology to compensate for the high-frequency loss of the channel, thereby improving the signal-to-noise ratio of the receiver signal from the source.

[0017] At the data link layer, this invention introduces channel coding and decoding technology based on soft information. By using Hamming code redundancy at the transmitting end and soft-decision Viterbi decoding at the receiving end, the original data is restored to the maximum extent possible using the probability information provided by the physical layer, thus achieving forward error correction. Finally, combined with a hybrid automatic repeat request mechanism, selective retransmission is triggered when the coding error correction capability is insufficient, constructing a closed-loop, adaptive noise-resistant communication system.

[0018] Through this cross-layer collaborative design, the present invention can significantly improve the communication rate and data integrity of digital isolators in harsh electromagnetic environments, and solve the contradiction between speed and reliability in the prior art.

[0019] To achieve the above objectives, the specific technical solution of the present invention is a high-speed data communication and noise suppression method for a digital isolator, comprising the following steps:

[0020] Step one: Data preprocessing and encoding at the transmitting end. The transmitting end performs channel coding on the data to be transmitted, generating encoded data containing redundancy check information. At the transmitting end, the original data is preprocessed, and an improved channel coding scheme based on Hamming codes is adopted. This step aims to add redundancy information to the original data, laying the foundation for error detection and correction at the receiving end, and resisting burst noise during transmission.

[0021] Step two involves noise estimation and pre-emphasis based on adaptive filtering. The coefficients of the pre-emphasis filter are adaptively adjusted according to channel characteristics to pre-emphasize the encoded data. Before the transmitting end drive circuit, the noise characteristics of the isolation channel (such as an on-chip transformer or capacitor) are estimated based on the feedback channel or a preset model, and the pre-emphasis intensity of the transmitted signal is dynamically adjusted. This step compensates for high-frequency attenuation in the channel, increasing the energy concentration of the signal in the transmission medium, thereby combating noise.

[0022] Step 3: Isolation transmission and dynamic threshold comparison. The pre-emphasized signal is transmitted through an isolator, and the encoded signal is transmitted through a digital isolator (magnetic or capacitive). The receiving end uses a dynamic threshold comparator instead of a traditional fixed threshold comparator. This step automatically adjusts the decision threshold by tracking the signal baseline drift and noise fluctuations in real time, which significantly reduces the bit error rate.

[0023] Step 4: Receiver soft decision decoding and error correction. The receiver samples and makes decisions on the received signal based on a dynamic threshold to obtain soft decision data containing signal amplitude information. The receiver synchronously samples the signal after passing through the isolator and comparator to obtain soft information that contains not only 0 / 1 hard decisions but also signal "confidence". Based on the soft information, the Viterbi algorithm is used to perform soft decision decoding on the convolutional code or the improved Hamming code to maximize the data recovery probability in noisy environments.

[0024] Step 5: Adaptive protocol retransmission and flow control. Soft decision data is used to perform soft decision decoding on the received signal to recover the original data. Combined with the error verification results after decoding, a hybrid automatic retransmission request mechanism is implemented. When decoding fails (the bit error rate exceeds the threshold), the receiver requests retransmission through the reverse isolation channel. At the same time, the transmitter dynamically adjusts the coding redundancy and transmission rate to achieve a balance between real-time performance and reliability.

[0025] Preferably, the channel coding uses Hamming code or extended Hamming code, and its coding process includes:

[0026] Define the original data block as ,in , The length of the information bits. Let be the generating matrix of k×n. This represents the total length of the codewords.

[0027] Based on the generator matrix The length is Information bits Mapped to a length of The code The encoded data codewords for:

[0028] ;

[0029] Here, the check bit is defined. The formula for calculating the check digit is:

[0030] ;

[0031] In the formula, For parity check matrix, i.e., k×m parity check matrix sub-blocks, This represents the XOR operation. Representation matrix The Middle Line number Column elements;

[0032] The generator matrix G and the parity check matrix H are used to construct a systematic linear block code, which is the basis for forward error correction. It ensures that after adding redundancy, the data can be constrained by a unique set of parity check equations at the receiving end, thereby locating and correcting errors.

[0033] This invention employs extended Hamming code, and its check bit length is... ,satisfy ,and (For the non-extended part), the Hamming code encoding model uses the original data vector. As the independent variable, the encoded codeword As the dependent variable, the Hamming code parameters (n,k) are selected to satisfy... , Construct the parity check matrix Its column vector is All non-zero vectors in the dimensional space;

[0034] Will Convert to system form ,in, As a unit array, for Matrix, and then construct a generating matrix. .

[0035] For system code, codeword According to the definition of Hamming code, it must satisfy... .

[0036] ;

[0037] Rearranging the terms, we get:

[0038] ;

[0039] Since addition is equivalent to XOR in the binary field, the parity bit... Through and Matrix number The expression obtained by performing a dot product (XOR sum) on the columns is: .

[0040] Preferably, the adaptive adjustment of the pre-emphasis filter coefficients includes:

[0041] Update the filter weight vector using the least mean square algorithm Its update formula (based on the LMS adaptive algorithm) is as follows:

[0042]

[0043] in, Step size factor For error signals, For the input signal, It is the filter weight vector when the codeword length is n. It is the filter weight vector when the code length is n+1;

[0044] The LMS (Learning Management System) adaptive algorithm is used to achieve adaptive tracking of time-varying channels and noise without prior knowledge of the precise channel model, enabling the system to self-optimize in complex electromagnetic environments.

[0045] The independent variable of the LMS adaptive algorithm is the input signal. And the expected response (reference signal) The dependent variable is the filter weights. ;

[0046] Define the filter output: ;

[0047] Define the error signal: ;

[0048] The goal is to minimize the mean squared error: .

[0049] Using the steepest descent method, the weight update formula is: In practice, instantaneous gradients are used. Approximately, substituting, we get:

[0050] ;

[0051] in, It is the step size factor that controls the convergence speed and steady-state error.

[0052] Frequency response of the pre-emphasis filter Configured to match channel frequency response They are inversely correlated, and the transmitted signal is defined as follows: The channel impulse response is The noise is Then the received signal is: ;

[0053] Preemphasis Filter The frequency response is designed to be the inverse function of the channel response, that is:

[0054] ;

[0055] In the formula, yes Fourier transform, To prevent positive numbers from being divided by zero, It is a very small positive number used to avoid division by zero and to limit high-frequency gain;

[0056] Preemphasis Filter Its function is to compensate for the channel. The attenuation of high-frequency components makes the signal waveform received at the receiving end more complete, and the eye diagram opens wider, which directly improves the signal-to-noise ratio.

[0057] Preferably, the received differential signal is defined as and , differential signal ;

[0058] The dynamic threshold is determined by the following formula:

[0059] ;

[0060] in, Forgetting factor, ( ), Within the current time window The median or average value, Let be the dynamic threshold at time t. The dynamic threshold at time t-1;

[0061] Dynamic threshold This invention solves the problem of inaccurate judgment under signal drift or noise interference in traditional fixed thresholds. By introducing historical states and current averages, it achieves smooth threshold updates and effectively suppresses false triggering caused by sudden noise.

[0062] Preferably, the sampling decision is based on a dynamic threshold and a preset hysteresis interval. When the differential signal is processed... Greater than The first output level is less than 1. The second level is output at the time; the comparator output... for:

[0063] ;

[0064] in, This is the hysteresis comparison interval, used to prevent frequent flipping caused by noise.

[0065] Preferably, the soft-decision data includes the log-likelihood ratio of the received signal sample values ​​as follows: ;

[0066] in, The sampled value of the received signal. For the corresponding original bits, This indicates that, given that the transmitted bit is 0, the receiving end samples the numerical value. The probability density, This indicates that, given that the transmitted bit is 1, the value obtained by the receiver through sampling... The probability density;

[0067] if Much larger ,but A large positive value indicates that the receiver is "confident" that 0 was sent;

[0068] if Larger, then A large negative value indicates that the sender is "confident" to be sending a 1;

[0069] If the two are close A value close to 0 indicates a very low credibility of the judgment. The Viterbi decoder will then combine the combined information from multiple symbols before and after the judgment to make a final decision.

[0070] Preferably, the soft-decision decoding employs the Viterbi algorithm, and its path metric update formula is:

[0071] ;

[0072] in, In time Arrival Status Path metrics In time Arrival Status The path metric, BM(s′→s), is the branch metric, calculated from the LLR of the received symbol and the likelihood of the corresponding output bit.

[0073] It should be noted that the log-likelihood ratio (LLR) converts the amplitude information of the physical layer signal into probability information. The soft-decision Viterbi algorithm uses this information for decoding. Compared with hard decision, soft decision can obtain 2-3 dB more coding gain, which is crucial for data recovery under weak signals.

[0074] The LMS weight update formula drives the dynamic adjustment of the adaptive pre-emphasis module, ensuring that the system always works in the optimal equilibrium state and adapts to changes in channel characteristics caused by differences in the manufacturing process or temperature of different batches of isolators.

[0075] Preferably, it also includes a hybrid automatic repeat request step:

[0076] The receiving end performs error checking on the data after soft-decision decoding;

[0077] If the verification fails, a retransmission request is sent to the sender through the reverse isolation channel;

[0078] The sending end dynamically adjusts the code rate of the channel coding based on the frequency of retransmission requests.

[0079] Preferably, the specific operation process of the entire method is as follows:

[0080] Transmitter startup and parameter initialization: After the system is powered on, the transmitter and receiver establish a communication link through a low-speed handshake signal. The transmitter is configured with an initial coding rate of 4 / 5 (Hamming code), and the pre-emphasis filter coefficients are set to the default values.

[0081] Data encoding: The sending end encodes the parallel data to be transmitted according to... One group is divided into blocks, according to the formula calculate Bit check bit, generated Bit code ;

[0082] Channel characteristic detection and pre-emphasis: During data transmission gaps, the transmitter sends a specific training sequence, and the receiver analyzes the attenuation of this sequence to calculate the error signal. The error information is then transmitted back via the reverse channel, and the transmitting end uses the LMS adaptive algorithm: Update the weights of the pre-emphasis filter to compensate for high-frequency components in the data to be transmitted;

[0083] Isolated transmission: A pre-emphasized digital signal drives the primary coil of an isolator (such as an on-chip transformer), transmitting the signal to the secondary side via magnetic field coupling. During this process, common-mode noise... Superimposed on the signal;

[0084] Receiver-side dynamic threshold processing: The receiver amplifier receives the differential signal. The threshold generation module generates the threshold according to the formula. The dynamic threshold is calculated in real time, and the comparator uses this threshold in conjunction with the hysteresis interval. Convert analog signals into digital sequences with soft information;

[0085] Soft decision decoding: The receiving end stores the sampled values. The LLR value for each bit is calculated. Then, the Viterbi decoder calculates the branch metric BM based on the LLR and accumulates, compares and selects the path metric PM on the grid graph. Finally, it backtracks to output the most likely data sequence.

[0086] Error control and feedback: The decoded data frame undergoes cyclic redundancy check. If the check fails, the receiver sends a NACK signal through the reverse isolation channel. After receiving the NACK, the transmitter initiates a retransmission mechanism and may reduce the coding rate (e.g., from 4 / 5 to 2 / 3) to improve noise immunity. If the check is successful continuously, the coding rate is gradually increased to improve throughput.

[0087] Data output: After being de-jittered, the correctly decoded data is output in parallel by the receiving end.

[0088] Preferably, a digital isolator uses the method described above for data communication.

[0089] Preferably, a control module for a digital isolator includes:

[0090] The transmitting end includes a channel coding module and an adaptive pre-emphasis filter;

[0091] The receiving end includes a dynamic threshold comparator and a soft-decision decoder;

[0092] The sending end and the receiving end cooperate to execute the steps of the method.

[0093] Preferably, a computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the method.

[0094] Compared with the prior art, the technical solution disclosed in this application has the following non-obvious technical features:

[0095] First, this application realizes the utilization of soft information through cross-layer collaboration. In the prior art, the analog signal of the physical layer loses amplitude information after passing through the comparator, and only retains 0 / 1 hard decision. This application breaks through this conventional practice, quantizes the analog amplitude of the physical layer into soft information (LLR), and directly provides it to the Viterbi decoder of the data link layer. This cross-layer design that transmits the confidence information of the physical layer to the link layer has not been publicly reported in the field of digital isolators and is not obvious.

[0096] Second, this application adopts joint optimization of dynamic threshold and adaptive pre-emphasis. Traditional schemes usually optimize the transmitter driver or receiver separately. This application not only adopts adaptive pre-emphasis and dynamic threshold respectively, but more importantly, it establishes a linkage mechanism between the two (such as collaborative calibration through training sequences), which solves the problem of pre-emphasis and threshold mismatch caused by channel nonlinearity and realizes joint optimization of transmitter and receiver.

[0097] Third, this application implements a coding rate adaptive function based on noise statistics. Existing retransmission mechanisms (such as ARQ) usually decide on retransmission based solely on the CRC check result (success or failure) of the data packet. This application further introduces LLR statistics (such as average likelihood ratio and number of error corrections) during the decoding process as a basis for judging channel quality, which can more sensitively perceive subtle changes in noise level, thereby making coding rate adjustment smoother and more accurate, avoiding the "roller coaster" rate jumps of traditional schemes.

[0098] Fourth, this application integrates hysteresis comparison and soft decoding. This application introduces a hysteresis interval (δ) in the dynamic threshold comparator. This hysteresis interval is not a fixed value, but is related to the error correction capability of the subsequent decoder. This design filters out the tiny jitters that the decoder can easily correct at the analog front end, reduces the amount of invalid computation at the digital back end, and achieves a power balance between noise immunity at the analog front end and error correction at the digital back end, which is inventive.

[0099] Compared with the prior art, the present invention has the following beneficial effects:

[0100] 1. This invention effectively improves the data transmission rate and reliability. By adaptive pre-emphasis to compensate for high-frequency attenuation of the channel and combined with dynamic threshold decision, it effectively reduces inter-symbol interference, so that the data transmission rate can be increased by more than 30% compared with the existing technology under the same isolation bandwidth. The introduction of soft decision decoding technology can obtain a coding gain of 2-3dB, which reduces the system bit error rate by more than an order of magnitude under the same signal-to-noise ratio.

[0101] 2. This invention enhances the ability to resist common-mode transient interference. The dynamic threshold technology can follow the baseline drift caused by common-mode noise in real time, instead of using a fixed window to shield noise. This greatly improves the system's ability to suppress high dv / dt (such as ±100kV / µs), avoids false triggering caused by common-mode noise, and ensures communication continuity in harsh industrial environments.

[0102] 3. This invention improves system adaptability: The adaptive algorithm based on LMS and the coding rate adjustment mechanism based on channel quality feedback enable this method to automatically adapt to manufacturing differences, operating voltage fluctuations and environmental temperature changes in different batches of isolators, without the need for manual calibration, thereby improving product yield and the universality of application scenarios.

[0103] 4. This invention optimizes the system's power consumption by dynamically adjusting the pre-emphasis intensity at the transmitter and the threshold window at the receiver, avoiding the use of maximum drive current under all operating conditions. When channel conditions are favorable, the transmission power can be reduced; when noise is low and decoding accuracy is high, the number of retransmissions can be reduced, thereby optimizing overall power consumption. Attached Figure Description

[0104] Figure 1 This is a flowchart of the method described in Embodiment 1 of the present invention;

[0105] Figure 2 This is a graph showing the data transmission rate of the digital isolator described in Embodiment 1 of the present invention;

[0106] Figure 3 This is a voltage curve of the differential signal of the digital isolator described in Embodiment 1 of the present invention;

[0107] Figure 4 This is a bar chart comparing the junction temperature of the digital isolator chip described in Embodiment 1 of the present invention;

[0108] Figure 5 This is a power consumption curve of the digital isolator described in Embodiment 1 of the present invention;

[0109] Figure 6 This is a schematic diagram of the control module described in Embodiment 2 of the present invention. Detailed Implementation

[0110] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings;

[0111] Example 1:

[0112] A high-speed data communication and noise suppression method using a digital isolator is disclosed. This method is applied to a communication system where the transmitting control logic is implemented using a Field-Programmable Gate Array (FPGA) and the receiving processing logic is implemented using an Application-Specific Integrated Circuit (ASIC). The isolation medium is an on-chip silicon dioxide capacitor. The method flow is as follows: Figure 1 As shown, it includes the following specific implementation process:

[0113] Step 101, Initialization and parameter configuration of the sending end:

[0114] After the system powers on, the transmitter and receiver establish a communication link via a low-speed handshake signal. The transmitter controller reads the preset initial configuration parameters: the channel coding method is configured as extended Hamming code (15,11), i.e., the information bit length... Length of check bits Total code length The initial coding rate is 11 / 15; the initial weight vector of the adaptive pre-emphasis filter. Let the step size be [0.2, 0.2, 0.2, 0.2, 0.2]. Set to 0.01; Forgetting factor for the receiver's dynamic threshold. Set to 0.875, hysteresis interval Set to 50mV;

[0115] Step 102, Data encoding at the sending end:

[0116] The sending end will transmit 11 bits of parallel data. to As information bits Based on the pre-generated verification matrix Calculate the 4-bit parity bit to The formula for calculating each check bit is as follows: For example, for the first parity bit By combining information bits with The first column of the matrix is ​​XORed to obtain the result. The parity bit is then appended to the information bits to form a 15-bit codeword.

[0117] .

[0118] Step 103, Channel characteristic detection and adaptive pre-emphasis coefficient:

[0119] During data transmission intervals (after every 1000 data frames), the transmitting end sends a known 32-bit training sequence. Upon receiving the training sequence, the receiving end compares it with a locally stored reference sequence and calculates the error signal. The error signal is transmitted back to the transmitter via a reverse isolation channel (using low-speed, high-reliability on / off keying modulation), and the transmitter updates the pre-emphasis filter weights according to the LMS algorithm: For example, if the current error signal The input signal is -0.05. If the value is 0.8, then the weight update amount is 0.01×(-0.05)×0.8=-0.0004. The new weight vector is obtained by subtracting this value from the original value, thereby fine-tuning the frequency response of the pre-emphasis filter so that it can more accurately compensate for the high-frequency attenuation of the channel.

[0120] Step 104, Pre-emphasis processing and isolated transmission:

[0121] The codewords of the encoded data to be transmitted at the sending end A parallel-to-serial conversion is performed to form a serial bit stream. This serial bit stream passes through a pre-emphasis filter, which pre-emphasizes the rising and falling edges of the signal, i.e., amplifies the high-frequency components of the signal according to the filter coefficients. The pre-emphasized signal drives the primary electrode of the on-chip capacitive isolator, and transmits the signal to the secondary side through electric field coupling. During this process, common-mode noise generated by the switching action of the motor drive circuit is introduced. ( Approximately 50 kV / µs) is superimposed on the signal;

[0122] Step 105, Dynamic threshold sampling and soft information generation at the receiving end:

[0123] The receiving amplifier receives the differential signal from the secondary side. The threshold generation module calculates dynamic thresholds in real time. ,in Within the current 10ns time window The average value, the comparator is based on the dynamic threshold and hysteresis interval. To make a judgment: If Output "1" and record the amplitude value at that sampling point. ;like Output "0" and record If it falls between these two states, the output retains the previous state; the receiver will output the sampling amplitude value corresponding to each bit. It is stored as an 8-bit soft information, which quantifies the confidence level of the decision;

[0124] Step 106, Soft Decision Decoding:

[0125] The receiver inputs the stored soft information sequence into the Viterbi decoder. First, the decoder analyzes the received signal... With the assumed transmitted bits Given a probability distribution of 0 or 1, calculate the log-likelihood ratio for each bit. Subsequently, the decoder runs the Viterbi algorithm on the trellis graph of the (15,11) extended Hamming code. For each time t and each state s, the path metric is computed. Branching metric Based on the accumulated LLR values ​​of the output bits corresponding to the transition, after 15 time steps, the decoder selects the surviving path with the maximum path metric and backtracks to output 11 original information bits.

[0126] Step 107, Hybrid Automatic Repeat and Adaptive Flow Control:

[0127] The decoded data frames undergo Cyclic Redundancy Check (CRC-16). If the CRC check passes, the receiver sends an ACK signal to the transmitter; if the check fails, the receiver sends a NACK signal through the reverse isolation channel, including a current channel quality indicator (based on the average absolute value of LLR during decoding). Upon receiving the NACK, the transmitter initiates a retransmission mechanism: if three consecutive retransmissions fail, the coding rate is reduced from 11 / 15 to 7 / 11 (i.e., redundancy is increased); if 100 consecutive frames are successful without NACK, the coding rate is gradually increased back to 11 / 15. Simultaneously, the transmitter dynamically adjusts the pre-emphasis strength based on the average value of the LLR feedback, forming a closed-loop optimization.

[0128] Step 108, Data Output:

[0129] The data frame that is correctly decoded and has passed CRC verification is then buffered and output by the receiving end in 11-bit parallel form to the subsequent processing unit, completing a full high-speed, high-noise-resistant data communication.

[0130] Figure 2 The data transmission rate of the digital isolator is demonstrated. The digital isolator in Example 1, under the combined effect of adaptive pre-emphasis, dynamic threshold and soft decoding, can achieve an initial rate of over 98Mbps. After a short training period, it can quickly increase to the 115-120Mbps range and can still maintain a high rate of over 115Mbps when encountering strong interference. In contrast, existing digital isolators are limited by fixed threshold and lack forward error correction. Their rate drops sharply in the interference environment and is difficult to recover to the initial level.

[0131] Figure 3 The voltage variation of the differential signal of the digital isolator is shown. The voltage of the differential signal reflects the actual usable signal amplitude at the receiver. The digital isolator in Example 1 compensates for high-frequency attenuation in the channel through adaptive pre-emphasis, which increases the differential voltage at the receiver to over 400mV after training and can still maintain over 390mV under interference and temperature changes. In contrast, existing digital isolators lack high-frequency compensation, and the signal continues to attenuate during transmission, with the final steady-state voltage being only about 322mV, significantly increasing the risk of eye diagram closure.

[0132] Figure 4 The paper demonstrates the junction temperature variation of a digital isolator chip. In Example 1, by dynamically adjusting the pre-emphasis intensity, optimizing the coding rate, and reducing invalid retransmissions, power consumption was significantly reduced, thereby controlling the rise in chip junction temperature. After 30 consecutive time points of operation, the junction temperature of the chip in this application stabilized at 39.5°C, while the junction temperature of the prior art reached 53.5°C due to continuous high power consumption operation, a temperature difference of 14°C. High temperature may lead to a shortened lifespan and decreased reliability of existing equipment.

[0133] Figure 5 The power consumption of the digital isolator is demonstrated. In Example 1, adaptive power consumption optimization is achieved by dynamically adjusting the pre-emphasis intensity at the transmitting end, reducing the number of retransmissions, and reducing redundancy when the channel is good. In the initial stage, the power consumption is slightly lower than that of the prior art. As the adaptive mechanism takes effect, the power consumption of this application further decreases and stabilizes in the range of about 25-28mW. In contrast, the existing digital isolators lack an adaptive mechanism, and their power consumption rises sharply to more than 45mW in the interference environment, and the recovery is slow. The final steady-state power consumption is still about 4.3mW higher than that of this application.

[0134] Example 2:

[0135] A control module for a digital isolator, the structure of which is as follows: Figure 6 As shown, the control module is configured to control a capacitive digital isolator. The control module includes a transmitter and a receiver, which are located on opposite sides of the isolation barrier. Its specific structure and implementation process are as follows:

[0136] Sender:

[0137] Its specific structure includes: an input buffer for receiving external data; a Hamming code encoder that internally stores a parity check matrix H; an LMS adaptive pre-emphasis filter containing tapped delay lines, multipliers, and accumulators; a training sequence storage ROM; an inverse decoder for parsing the inverse channel signal; and a driver for driving the isolation capacitor.

[0138] The specific implementation process involves the input buffer sending the received data to the Hamming code encoder, which then executes the data according to the formula... The parity bit is calculated in real time, and the encoded data is output to the LMS adaptive pre-emphasis filter. The LMS filter calculates the parity bit according to the error signal e(n) provided by the inverse decoder. The weights are updated. When in training mode, the data in the training sequence ROM is gated and sent to the filter. The filtered signal is amplified by the driver and then applied to the primary plate of the isolation capacitor.

[0139] Receiver:

[0140] Its specific structure includes: a transimpedance amplifier for converting the current of the secondary plate of the isolation capacitor into a voltage; a dynamic threshold calculation unit containing a sliding window averager and a multiplier; a hysteresis comparator whose threshold is provided by the dynamic threshold calculation unit; an analog-to-digital converter for sampling the amplitude of the input signal when the comparator output transitions; a Viterbi decoder containing a branch metric calculation unit, an adder-compare-select unit, and a backtracking unit; a CRC checker; and a reverse encoder for sending control signals.

[0141] The specific implementation process involves the transimpedance amplifier receiving the current signal from the secondary plate of the isolation capacitor and converting it into a voltage. The dynamic threshold calculation unit samples at fixed time intervals. Calculate the average value within the window. and according to Generate dynamic threshold The hysteresis comparator will and The comparison is performed, and an initial digital decision is output. Simultaneously, the analog-to-digital converter performs a comparison at the decision point. Sampling is performed to generate a 4-bit soft information sequence (quantization precision of 16 levels). The Viterbi decoder receives this soft information sequence, first calculates the LLR, and then performs the add-compare-select operation: Finally, the original data is back-tracked and output. The CRC checker verifies the decoded original data. If successful, it is output through the output buffer; if it fails, the reverse encoder is triggered, and a retransmission request is sent to the feedback decoder at the transmitting end through the reverse channel (using the secondary side of the same isolation capacitor to send a signal to the primary side).

[0142] Through the coordinated operation of the aforementioned transmitting and receiving ends, this control module achieves high-speed, noise-resistant control of the capacitive digital isolator.

[0143] Example 3:

[0144] A digital isolator is provided, which has the control function of the control module in Embodiment 2. The digital isolator adopts a dual-chip stacked package form, and its internal components integrate a transmitter chip and a receiver chip in accordance with the control module in Embodiment 2. The two chips are magnetically coupled and isolated from each other through an on-chip transformer.

[0145] The transmitting chip integrates:

[0146] A channel coding module for performing extended Hamming code encoding;

[0147] An adaptive pre-emphasis filter module whose coefficients are dynamically updated by the built-in LMS engine based on the reverse channel feedback;

[0148] A training sequence generator is used to periodically send known sequences to probe the channel;

[0149] A reverse channel receiver is used to receive error signals and retransmission requests from the receiving chip;

[0150] The receiver chip integrates:

[0151] A dynamic threshold comparator module is used to dynamically adjust the decision threshold based on the median value of the real-time signal;

[0152] A soft information extraction module is used to quantize the sampling amplitude;

[0153] A soft-decision decoder module (Viterbi decoder) is used to recover the original data using soft information;

[0154] A CRC check and ARQ controller is used to generate ACK / NACK signals;

[0155] A reverse channel transmitter is used to send control signals back to the transmitting chip.

[0156] In practical implementation, the transmitting chip of the digital isolator receives 11-bit parallel data from the microcontroller through its input pin; the channel coding module immediately encodes it into a 15-bit codeword and sends it to the adaptive pre-emphasis filter; the pre-emphasis filter performs high-frequency compensation on the code stream based on the weights updated in the previous channel detection cycle; the compensated signal drives the primary coil of the on-chip transformer to generate a changing magnetic field; the secondary coil of the receiving chip senses the change in the magnetic field and generates an induced voltage; the dynamic threshold comparator tracks the baseline drift of this induced voltage in real time and converts it into a digital sequence with amplitude information according to the dynamic threshold; this sequence is used as soft information input to the Viterbi decoder; the decoder performs path metric calculation and backtracking, outputs 11-bit raw data, and sends it to the microcontroller through its output pin; throughout the process, if the CRC check fails, the ARQ controller sends a retransmission request to the transmitting chip through the reverse channel transmitter; after receiving the request, the reverse channel receiver of the transmitting chip controls the channel coding module to retransmit the previous frame of data.

[0157] This digital isolator, through its internally integrated functional modules, automatically completes the entire process from data input, encoding, pre-emphasis, isolated transmission, dynamic threshold decision, soft decoding to retransmission control without the intervention of an external controller, achieving reliable communication at data rates up to 150 Mbps and common-mode transient immunity of ±150 kV / µs.

[0158] Example 4:

[0159] A computer-readable storage medium is a non-volatile flash memory embedded in a digital isolator transmitter chip. The memory stores a computer program (i.e., firmware code) that, when executed by an embedded processor (or hardwired state machine) within the transmitter and receiver chips, performs the following steps:

[0160] When the computer program is executed on the processor of the transmitting chip, it achieves:

[0161] Encoding control steps: Read the data buffer to be sent, calculate the parity bit using the preset generator matrix G according to the rules of extended Hamming code (15,11), and generate the encoded data stream;

[0162] Training and Adaptation Steps: The training sequence is sent periodically. Upon receiving the error signal from the receiving chip, the LMS algorithm function is called and executed. Update the preemphasis filter coefficients stored in the register;

[0163] Pre-emphasis filtering step: Input the encoded data stream into the digital filter, use the updated values ​​of the filter coefficients from the previous step, perform convolution operation on the data stream to achieve pre-emphasis of high-frequency components;

[0164] Retransmission control steps: Monitor the reverse channel. When a NACK signal is received, backtrack the transmit buffer pointer to the previous frame of data, retransmit the frame, and record the number of retransmissions. Adjust the coding rate parameter according to the retransmission frequency (e.g., switch from 11 / 15 to 7 / 11).

[0165] When the computer program is executed on the processor of the receiving chip, it achieves:

[0166] Dynamic threshold calculation steps: Sample the differential voltage of the received signal with a period of 1 ns, and calculate the moving average of the most recent N sampling points. and execute Recursive operations;

[0167] Soft decision sampling step: Based on the transition edge of the comparator output, the analog-to-digital converter is triggered to sample the current signal amplitude, and the amplitude value is stored as a soft bit in the soft information buffer;

[0168] The Viterbi decoding process involves: reading the sequence from the soft information buffer and calculating the LLR value for each received symbol; then executing the core loop of the Viterbi algorithm: initializing the path metric, calculating the branch metric for each state, performing an add-compare-select operation to update the path metric, and recording surviving paths; and finally, performing a backtracking operation at the end of the decoding window to output the data after the decision.

[0169] Error control steps: Perform CRC check on the decoded data. If successful, send ACK and output the data. If unsuccessful, send NACK and encode the average LLR value (as a channel quality indicator) of the current decoding process and send it through the reverse channel.

[0170] The program stored on the computer-readable storage medium, through the execution of the above steps, enables the digital isolator to possess intelligent communication capabilities of self-adaptation, self-correction, and self-optimization, thus realizing the technical solution defined in this invention.

[0171] The above technical solutions only embody the preferred technical solutions of the present invention. Any modifications that may be made by those skilled in the art to certain parts thereof embody the principles of the present invention and fall within the protection scope of the present invention.

Claims

1. A method for high-speed data communication and noise suppression using a digital isolator, characterized in that, Includes the following steps: Step 1: The transmitting end performs channel coding on the data to be transmitted to generate coded data containing redundancy check information; Step 2: Adaptively adjust the coefficients of the pre-emphasis filter according to the channel characteristics to pre-emphasize the encoded data; Step 3: Transmit the pre-emphasized signal through the isolator; Step 4: The receiving end samples and makes decisions on the received signal based on a dynamic threshold to obtain soft decision data containing signal amplitude information. Step 5: Use the soft decision data to perform soft decision decoding on the received signal to recover the original data; The soft-decision data includes the log-likelihood ratio of the received signal sample values: ; in, The sampled value of the received signal. For the corresponding original bits, This indicates that, given that the transmitted bit is 0, the receiving end samples the numerical value. The probability density, This indicates that, given that the transmitted bit is 1, the value obtained by the receiver through sampling... The probability density; The soft-decision decoding uses the Viterbi algorithm, and its path metric update formula is as follows: ； Among them, branching measure It is calculated based on the log-likelihood ratio.

2. The method according to claim 1, characterized in that, The channel coding employs Hamming code or extended Hamming code, and its coding process includes: Based on the generator matrix The length is Information bits Mapped to a length of The code ,satisfy ; The formula for generating the check digit is as follows: , For the verification matrix, This represents the XOR operation.

3. The method according to claim 1, characterized in that, The adaptive adjustment of the pre-emphasis filter coefficients includes: Update the filter weight vector using the least mean square algorithm Its update formula is: ； in, Step size factor For error signals, For the input signal, It is the filter weight vector when the codeword length is n. It is the filter weight vector when the code length is n+1; Frequency response of the pre-emphasis filter Configured to match channel frequency response They are inversely correlated, that is ϵ is a positive number to prevent division by zero.

4. The method according to claim 1, characterized in that, The dynamic threshold is determined by the following formula: ； in, Forgetting factor, This refers to the median or average value of the received signal within the current time window. Let be the dynamic threshold at time t. The dynamic threshold at time t-1; The sampling decision is based on a dynamic threshold and a preset hysteresis interval. When the differential signal is processed... Greater than The first level is output when the differential signal is active. Less than The second level is output at that time.

5. The method according to claim 1, characterized in that, It also includes the Hybrid Automatic Repeat Request step: The receiving end performs error checking on the data after soft-decision decoding; If the verification fails, a retransmission request is sent to the sender through the reverse isolation channel; The sending end dynamically adjusts the code rate of the channel coding based on the frequency of retransmission requests.

6. A control module for a digital isolator, characterized in that, include: The transmitting end includes a channel coding module and an adaptive pre-emphasis filter; The receiving end includes a dynamic threshold comparator and a soft-decision decoder; The transmitting end and the receiving end cooperate to perform the steps of the method as described in any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 5.

Images