A method for enhancing the transient response of voltage regulators based on load change rate feedforward prediction

By acquiring the load current change rate and determining the switching direction in real time, dynamically adjusting the compensation intensity and duration, and using a charging and discharging array for precise compensation, the problem of LDO response lag and high power consumption when the load current changes is solved, and a fast, low-power transient response of the regulator is achieved.

CN123067652APending Publication Date: 2026-11-10FUZHOU YILI ELECTRIC POWER ENG CO LTD +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611305014.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-26
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing low dropout linear regulators (LDOs) cannot respond promptly to sudden changes in load current, resulting in undershoot or overshoot of the output voltage. Furthermore, their compensation strength is fixed and they cannot provide differentiated responses, leading to high power consumption.

Method used

By collecting the load current change rate in real time, the direction and severity of load jumps are determined, and the compensation intensity, direction and duration are dynamically adjusted. Precise compensation is performed using a charge and discharge array, and parameters are optimized through comprehensive scoring.

Benefits of technology

It enables early identification and direction determination of load transitions, differentiated response, reduced static power consumption, improved response speed and consistency, optimized compensation parameters, and adaptability to different load conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123067652A_ABST
    Figure CN123067652A_ABST
Patent Text Reader

Abstract

This application relates to the field of power management technology, and in particular to a method for enhancing the transient response of a voltage regulator based on load rate of change feedforward prediction. The method includes: real-time acquisition of load current and determination of its rate of change, thereby determining load jumps and their directions; responding to load jumps, determining compensation intensity based on the ratio of the rate of change amplitude to a preset maximum rate of change, determining pull-down or pull-up compensation direction based on the jump direction, and determining compensation duration based on the ratio of the rate of change amplitude to the expected current change; determining the corresponding number of units in the charging / discharging array, outputting compensation current to the power transistor control terminal according to the compensation direction, and controlling the duration according to the compensation duration; acquiring recovery effect indicators, adjusting the total number of units and unit current values ​​in the charging / discharging array accordingly, and using the adjusted parameters for subsequent compensation control. This application improves the transient response performance of the voltage regulator by detecting the load current rate of change and performing feedforward prediction compensation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of power management technology, and in particular to a method for enhancing the transient response of a voltage regulator based on load change rate feedforward prediction. Background Technology

[0002] Low-dropout linear regulators (LDOs) are a core component of power management integrated circuits and are widely used in scenarios with stringent power quality requirements. When the load current undergoes a step change, the gate voltage of the power transistor cannot keep up with the output change of the error amplifier due to the limited feedback loop bandwidth, resulting in undershoot or overshoot of the output voltage. Existing transient response enhancement technologies are mainly divided into the following categories: one is transient enhancement circuits, which detect voltage jumps when the output voltage changes abruptly and charge and discharge the gate of the power transistor, but this scheme must wait until the output voltage deviates from the target value before it starts to act, resulting in an inherent delay in response; another is adaptive bias technology, which dynamically adjusts the bias current of the error amplifier to improve the loop response speed, but cannot actively intervene before load changes occur; and yet another is load current sensing technology, which adjusts the output current by sensing load changes, but current sensing itself takes time, and voltage deviation has already occurred.

[0003] Existing solutions also have problems such as fixed compensation intensity, inability to differentiate responses based on transient severity, inability to adaptively optimize compensation parameters in a closed loop, and high power consumption due to the need to continuously consume static current to maintain a fast response. Summary of the Invention

[0004] To address these issues, this invention provides a method for enhancing the transient response of a voltage regulator based on load change rate feedforward prediction. This method overcomes the problems in existing technologies, such as fixed compensation intensity, inability to differentiate responses based on transient severity, inability to adaptively optimize compensation parameters in a closed-loop manner, and high power consumption due to the continuous consumption of static current to maintain a fast response.

[0005] To achieve the above objectives, this invention provides a method for enhancing the transient response of a voltage regulator based on load change rate feedforward prediction. It includes:

[0006] Step S1: Real-time acquisition of load current and determination of the rate of change of load current; based on the rate of change, determination of whether a load jump occurs and the direction of the load jump.

[0007] Step S2: In response to a load change, determine the compensation intensity based on the ratio of the magnitude of the rate of change to the preset maximum rate of change, determine the compensation direction as pull-down compensation or pull-up compensation based on the direction of the load change, and determine the compensation duration based on the ratio of the magnitude of the rate of change to the expected current change.

[0008] Step S3: Determine the number of charge and discharge units to be selected according to the compensation intensity index, select the corresponding number of charge and discharge units in the charge and discharge array composed of multiple charge and discharge units according to the compensation direction, output compensation current to the power tube control terminal of the voltage regulator, and control the compensation duration according to the compensation duration.

[0009] Step S4: Collect the recovery effect index for each load transition event, adjust the total number of units in the charging and discharging array and the current value of a single charging and discharging unit according to the recovery effect index, and use the adjusted parameters for compensation control in subsequent load transition events.

[0010] Further, in step S1, determining the rate of change of the load current includes:

[0011] The load current is continuously sampled at a fixed sampling frequency to obtain a digital sequence of the load current;

[0012] The original value of the rate of change is calculated by dividing the difference between the digital load current at the current moment and the previous moment by the sampling interval. The calculation formula is as follows:

[0013]

[0014] in, The original value of the rate of change. This is the digital value of the load current at the current moment. This represents the digital value of the load current at the previous moment. The sampling interval;

[0015] The original value of the rate of change is subjected to digital low-pass filtering to obtain the filtered rate of change, and the calculation formula is as follows:

[0016]

[0017] in, The filtered rate of change at the current time. These are the filter coefficients. This represents the rate of change after filtering at the previous moment.

[0018] Further, in step S1, determining whether a load jump has occurred and the direction of the load jump based on the rate of change includes:

[0019] The rate of change is compared with preset positive and negative thresholds;

[0020] If the rate of change is greater than the positive threshold, it is determined to be a rising edge load jump;

[0021] If the rate of change is less than the negative threshold, it is determined to be a falling edge load jump;

[0022] If the absolute value of the rate of change is less than the positive threshold, it is determined that no load jump has occurred.

[0023] Further, in step S2, the formula for calculating the compensation strength index is:

[0024]

[0025] Where P is the compensation strength, This is the preset maximum rate of change.

[0026] Further, in step S2, determining the compensation direction as pull-down compensation or pull-up compensation based on the load switching direction includes:

[0027] If the load jumps on the rising edge, the compensation direction is pull-down compensation, which is used to pull down the control terminal of the power transistor for discharge.

[0028] If the load transition occurs on a falling edge, the compensation direction is pull-up compensation, which is used to pull up and charge the control terminal of the power transistor.

[0029] Further, in step S2, the formula for calculating the compensation duration is:

[0030]

[0031] in, To compensate for the duration, To preset the maximum compensation time, This represents the expected change in current.

[0032] Further, in step S3, the charging and discharging array includes a pull-down compensation branch and a pull-up compensation branch. The pull-down compensation branch contains N parallel pull-down discharging units, and the pull-up compensation branch contains M parallel pull-up charging units. The number of selected units is determined according to the compensation intensity index, including:

[0033] The number of gated units is determined based on the compensation intensity index, and the calculation formula is as follows:

[0034]

[0035]

[0036] Where P is the compensation intensity index, N is the total number of units in the pull-down compensation branch, and M is the total number of units in the pull-up compensation branch. To determine the number of units selected for the pull-down compensation branch, The number of units selected for the pull-up compensation branch;

[0037] If the load transition occurs on a rising edge, then N of the pull-down compensation branches will be selected. on Each pull-down discharge unit injects pull-down compensation current into the control terminal of the power transistor, and the total pull-down current is:

[0038]

[0039] in, For the total pull-down current, This is the selection control bit for the i-th pull-down discharge unit. The discharge current of a single pull-down discharge unit;

[0040] If the load transition occurs on a falling edge, then the number of pull-up compensation branches selected is [number missing]. Each pull-up charging unit injects pull-up compensation current into the control terminal of the power transistor, and the total pull-up current is:

[0041]

[0042] in, For the total pull-up current, This is the strobe control bit for the j-th pull-up charging unit. The charging current for a single pull-up charging unit;

[0043] The pull-down compensation current or pull-up compensation current is superimposed on the output current of the error amplifier in the voltage regulator at the power transistor control node.

[0044] Further, in step S4, recovery performance metrics for each load shift event are collected, including:

[0045] The actual recovery time of the load jump event is collected, which is the length of time from the moment when the amplitude of the rate of change first exceeds the preset threshold until the output voltage recovers to near the target value and continues to remain for more than the preset stable time.

[0046] The maximum voltage deviation of the load jump event is collected, and the absolute value of the maximum undershoot or overshoot of the output voltage during the recovery process is obtained.

[0047] Further, in step S4, adjusting the total number of cells in the charge-discharge array and the current value of a single charge-discharge cell according to the recovery effect index includes:

[0048] A comprehensive score is calculated based on the ratio of the actual recovery time to the target recovery time and the ratio of the maximum voltage deviation to the target voltage deviation. The calculation formula is as follows:

[0049]

[0050] Where Q represents the overall score. , The weighting coefficients and To preset the target recovery time, This refers to the actual recovery time. To prevent deviation from the preset target voltage, This represents the maximum voltage deviation.

[0051] When the overall score is less than the preset score, the total number of units in the charge-discharge array and the current value of a single charge-discharge unit are adjusted, and the adjusted parameters are stored in the on-chip non-volatile memory.

[0052] Further, adjusting the total number of cells in the charge / discharge array and the current value of a single charge / discharge cell includes:

[0053] When the comprehensive score of a series of transient events is less than the preset threshold and the compensation intensity index is in the saturation region, the total number of units in the charge-discharge array is increased.

[0054] When the comprehensive score of a series of transient events is less than the preset threshold and the compensation intensity index is in the low value range, the total number of units in the charge-discharge array is reduced.

[0055] When the negative impact of the actual recovery time on the overall score is greater than the maximum voltage deviation, the current value of a single charge / discharge unit is increased;

[0056] When the negative impact of the maximum voltage deviation on the overall score is greater than the actual recovery time, the current value of a single charge / discharge unit is reduced.

[0057] Each adjustment should not exceed a preset percentage of the original value.

[0058] Compared with existing technologies, the advantages of this invention are as follows: This embodiment acquires the load current in real time through a power transistor-based proportional sampling method and performs continuous sampling at a fixed sampling frequency. Combined with digital differentiation and low-pass filtering, the filtered rate of change can be output within an extremely short delay. By comparing this rate of change with positive and negative thresholds, the load transition and its direction can be determined before the output voltage deviates significantly from the target value. When the rate of change is greater than the positive threshold or less than the negative threshold, it is determined as a rising edge or falling edge load transition, respectively. This effectively achieves early identification of transients and accurate determination of direction, providing a reliable triggering basis for the rapid initiation of subsequent feedforward compensation and avoiding the response lag problem caused by waiting for the output voltage to deviate before triggering compensation in traditional solutions. At the same time, low-pass filtering eliminates false triggering caused by high-frequency noise, ensuring the reliability of detection.

[0059] Furthermore, this embodiment quantifies the severity of transients through a compensation intensity index, achieving differentiated responses in compensation intensity. The compensation intensity is small during light-load transients and large during heavy-load transients, avoiding the insufficient or excessive compensation caused by the "one-size-fits-all" approach of traditional solutions. Through nonlinear mapping processing, the risk of noise-induced false triggering is eliminated while maintaining sensitivity for small-signal transient identification. Linearity is maintained near the saturation region, ensuring consistent response under high-intensity transients. The compensation duration automatically shortens as the rate of change amplitude increases and automatically extends as the transient duration lengthens, ensuring the compensation duration matches the actual transient duration. The compensation direction is adaptively selected based on the transient direction, precisely suppressing undershoot and overshoot respectively.

[0060] Furthermore, this embodiment achieves continuously adjustable compensation intensity through a thermometer code gating method, avoiding current glitches during the gating process. The pull-down compensation branch and pull-up compensation branch are controlled independently, providing precise compensation for rising and falling edge transients respectively. The compensation current and the error amplifier output current are linearly superimposed at the gate node, which does not affect the normal operation of the main loop and can provide transient compensation current far exceeding the static drive capability, effectively solving the contradiction between response speed and static power consumption in traditional solutions. At the same time, in steady state without transient events, the gating control bits of all charging and discharging units are zero, the current drawn by the compensation array from the power rail is zero, and only nanoampere-level leakage current exists, fundamentally achieving static zero power consumption.

[0061] Furthermore, this embodiment quantifies the transient recovery effect of each time through comprehensive scoring and automatically fine-tunes the compensation parameters accordingly, achieving closed-loop adaptive optimization. The fine-tuning strategy is differentiated for both the total number of units and the unit current, based on the range of the compensation intensity index and the type of the deductible term, enabling the system to adapt to the optimal compensation requirements under different load conditions. The adjustment range is limited each time, ensuring the stability of the optimization process. The establishment of a historical optimal parameter library and a warm-up mechanism allow new transient events to directly call the matching optimal parameters without having to start from the default values ​​and gradually optimize, significantly improving the system's convergence speed and long-term adaptability.

[0062] Furthermore, this embodiment establishes a historical optimal parameter library and a warm-up mechanism, enabling new transient events to directly call the matching optimal parameters without having to gradually optimize from the default values, significantly improving the system's convergence speed. Through the adaptive update and condition reconstruction mechanism of the parameter library, it ensures that the parameter library remains effective over the long term, following the evolution of load characteristics and avoiding the problem of historical parameters becoming invalid due to fundamental changes in load characteristics. The step-by-step shutdown strategy avoids gate voltage rebound caused by instantaneous complete shutdown, ensuring a smooth and stable compensation termination process. By turning off all charging and discharging units when there are no transients, static zero power consumption is achieved, fundamentally solving the problem of consuming large static current to maintain fast response capabilities in traditional solutions. Attached Figure Description

[0063] Figure 1 This is a flowchart of the voltage regulator transient response enhancement method based on load change rate feedforward prediction according to an embodiment of this application;

[0064] Figure 2 This is a flowchart illustrating the determination of the load jump direction in the voltage regulator transient response enhancement method based on load change rate feedforward prediction in the embodiments of this application. Detailed Implementation

[0065] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0066] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0067] Please see Figures 1-2 As shown, Figure 1 This is a flowchart of the voltage regulator transient response enhancement method based on load change rate feedforward prediction according to an embodiment of this application; Figure 2 This is a flowchart illustrating the determination of the load jump direction in the voltage regulator transient response enhancement method based on load change rate feedforward prediction in the embodiments of this application.

[0068] The technical solution provided in this application includes the following steps:

[0069] Step S1: Real-time acquisition of load current and determination of the rate of change of load current; based on the rate of change, determination of whether a load jump occurs and the direction of the load jump.

[0070] Step S2: In response to a load change, determine the compensation intensity based on the ratio of the magnitude of the rate of change to the preset maximum rate of change, determine the compensation direction as pull-down compensation or pull-up compensation based on the direction of the load change, and determine the compensation duration based on the ratio of the magnitude of the rate of change to the expected current change.

[0071] Step S3: Determine the number of charge and discharge units to be selected according to the compensation intensity index, select the corresponding number of charge and discharge units in the charge and discharge array composed of multiple charge and discharge units according to the compensation direction, output compensation current to the power tube control terminal of the voltage regulator, and control the compensation duration according to the compensation duration.

[0072] Step S4: Collect the recovery effect index for each load transition event, adjust the total number of units in the charging and discharging array and the current value of a single charging and discharging unit according to the recovery effect index, and use the adjusted parameters for compensation control in subsequent load transition events.

[0073] Specifically, in step S1, determining the rate of change of the load current includes:

[0074] The load current is continuously sampled at a fixed sampling frequency to obtain a digital sequence of the load current;

[0075] The original value of the rate of change is calculated by dividing the difference between the digital load current at the current moment and the previous moment by the sampling interval. The calculation formula is as follows:

[0076]

[0077] in, The original value of the rate of change. This is the digital value of the load current at the current moment. This represents the digital value of the load current at the previous moment. The sampling interval;

[0078] The original value of the rate of change is subjected to digital low-pass filtering to obtain the filtered rate of change, and the calculation formula is as follows:

[0079]

[0080] in, The filtered rate of change at the current time. These are the filter coefficients. This represents the rate of change after filtering at the previous moment.

[0081] In this embodiment, the output power transistor of the LDO regulator is a PMOS transistor, with its source connected to the input power supply and its drain serving as the output to supply power to the load. A current detection circuit is set in the output branch of the power transistor, employing a power transistor replication proportional sampling method. The power transistor is connected in parallel with a detection transistor of proportional size, the size of which is 1 / 500 of the power transistor. The drain current of the detection transistor is equal to 1 / 500 of the load current, and this current is converted into a voltage signal through a sensing resistor. The voltage signal is continuously sampled at a sampling frequency of 200MHz (i.e., a sampling interval of 5ns), and after analog-to-digital conversion, a digital sequence of the load current is obtained, denoted as . ...The raw value of the rate of change is calculated by dividing the difference between the digital load current at the current moment and the previous moment by the sampling interval. This raw value is then subjected to a digital low-pass filter to eliminate false triggering caused by high-frequency noise. In this embodiment, a first-order IIR digital low-pass filter is used, with a filter coefficient α of 0.4.

[0082] Specifically, in step S1, determining whether a load jump has occurred and the direction of the load jump based on the rate of change includes:

[0083] The rate of change is compared with preset positive and negative thresholds;

[0084] If the rate of change is greater than the positive threshold, it is determined to be a rising edge load jump;

[0085] If the rate of change is less than the negative threshold, it is determined to be a falling edge load jump;

[0086] If the absolute value of the rate of change is less than the positive threshold, it is determined that no load jump has occurred.

[0087] In this embodiment, the filtered rate of change is compared with preset positive and negative thresholds. The absolute values ​​of the positive and negative thresholds are equal, both being 10 mA / ns. If the filtered rate of change is greater than the positive threshold, it indicates that the load current is rapidly increasing, and the load is transitioning from a light load to a heavy load. This is determined to be a rising edge load transition, which will cause the output voltage to undershoot. If the filtered rate of change is less than the negative threshold, it indicates that the load current is rapidly decreasing, and the load is transitioning from a heavy load to a light load. This is determined to be a falling edge load transition, which will cause the output voltage to overshoot. If the absolute value of the filtered rate of change is not greater than the preset threshold, it indicates that the load current is changing smoothly or only fluctuates slightly. This is determined to be no significant load transition, and no subsequent compensation action is triggered.

[0088] This embodiment acquires the load current in real time using a power transistor-based proportional sampling method and performs continuous sampling at a fixed sampling frequency. Combined with digital differentiation and low-pass filtering, the filtered rate of change is output within an extremely short delay. By comparing this rate of change with positive and negative thresholds, the load transition and its direction can be determined before the output voltage significantly deviates from the target value. When the rate of change is greater than the positive threshold or less than the negative threshold, it is determined as a rising edge or falling edge load transition, respectively. This effectively achieves early identification of transients and accurate direction determination, providing a reliable triggering basis for the rapid initiation of subsequent feedforward compensation. This avoids the response lag problem caused by waiting for the output voltage to deviate before triggering compensation in traditional solutions. Simultaneously, low-pass filtering eliminates false triggers caused by high-frequency noise, ensuring the reliability of the detection.

[0089] Specifically, in step S2, the formula for calculating the compensation strength index is:

[0090]

[0091] Where P is the compensation strength, This is the preset maximum rate of change.

[0092] Specifically, in step S2, determining the compensation direction as pull-down compensation or pull-up compensation based on the load switching direction includes:

[0093] If the load jumps on the rising edge, the compensation direction is pull-down compensation, which is used to pull down the control terminal of the power transistor for discharge.

[0094] If the load transition occurs on a falling edge, the compensation direction is pull-up compensation, which is used to pull up and charge the control terminal of the power transistor.

[0095] Specifically, in step S2, the formula for calculating the compensation duration is:

[0096]

[0097] in, To compensate for the duration, To preset the maximum compensation time, This represents the expected change in current.

[0098] In this embodiment, the absolute value of the filtered rate of change output in step S1 is used as the rate of change amplitude. The preset maximum rate of change is 1 A / ns. The ratio of the rate of change amplitude to the preset maximum rate of change is multiplied by 100 and rounded to obtain a compensation intensity index between 0 and 100. If the ratio exceeds 100, it saturates to 100. Based on this, a nonlinear mapping process is performed on the compensation intensity index. Values ​​less than 5 are set to zero as a dead zone to avoid false triggering by noise; values ​​greater than 95 are saturated to 100 to maintain a full response to high-intensity transients; values ​​between these two values ​​maintain a linear output, resulting in the mapped compensation intensity index.

[0099] The compensation direction is determined based on the load transition direction determined in step S1. If the load transition is on the rising edge, i.e., when the load transitions from light load to heavy load, the compensation direction is pull-down compensation, which is used to pull down the control terminal of the power transistor to discharge, thereby accelerating the turn-on of the power transistor and quickly supplying the increased load current. If the load transition is on the falling edge, i.e., when the load transitions from heavy load to light load, the compensation direction is pull-up compensation, which is used to pull up the control terminal of the power transistor to charge, thereby accelerating the turn-off of the power transistor and suppressing output voltage overshoot.

[0100] The preset maximum compensation time is 200ns, and the expected current change is 100mA. The initial estimated compensation duration is obtained by dividing the expected current change by the absolute value of the filtered rate of change. If this estimated value exceeds the preset maximum compensation time, the preset maximum compensation time is used as the final compensation duration; otherwise, the estimated value is used as the final compensation duration. When the filtered rate of change consistently exceeds the threshold, the compensation duration is automatically extended; when the filtered rate of change falls below the threshold, compensation terminates prematurely.

[0101] This embodiment quantifies the severity of transients through a compensation intensity index, achieving differentiated compensation intensity responses. The compensation intensity is small during light-load transients and large during heavy-load transients, avoiding the insufficient or excessive compensation caused by the "one-size-fits-all" approach of traditional solutions. Through nonlinear mapping processing, the risk of noise-induced false triggering is eliminated while maintaining sensitivity for small-signal transient identification. Linearity is maintained near the saturation region, ensuring consistent response under high-intensity transients. The compensation duration automatically shortens as the rate of change amplitude increases and automatically extends as the transient duration lengthens, ensuring the compensation duration matches the actual transient duration. The compensation direction is adaptively selected based on the transient direction, precisely suppressing undershoot and overshoot respectively.

[0102] Specifically, in step S3, the charging and discharging array includes a pull-down compensation branch and a pull-up compensation branch. The pull-down compensation branch contains N parallel pull-down discharging units, and the pull-up compensation branch contains M parallel pull-up charging units. The number of selected units is determined according to the compensation strength index, including:

[0103] The number of gated units is determined based on the compensation intensity index, and the calculation formula is as follows:

[0104]

[0105]

[0106] Where P is the compensation intensity index, N is the total number of units in the pull-down compensation branch, and M is the total number of units in the pull-up compensation branch. To determine the number of units selected for the pull-down compensation branch, The number of units selected for the pull-up compensation branch;

[0107] If the load transition occurs on a rising edge, then the number of pull-down compensation branches selected is [number missing]. Each pull-down discharge unit injects pull-down compensation current into the control terminal of the power transistor, and the total pull-down current is:

[0108]

[0109] in, For the total pull-down current, This is the selection control bit for the i-th pull-down discharge unit. The discharge current of a single pull-down discharge unit;

[0110] If the load transition occurs on a falling edge, then the number of pull-up compensation branches selected is [number missing]. Each pull-up charging unit injects pull-up compensation current into the control terminal of the power transistor, and the total pull-up current is:

[0111]

[0112] in, For the total pull-up current, This is the strobe control bit for the j-th pull-up charging unit. The charging current for a single pull-up charging unit;

[0113] The pull-down compensation current or pull-up compensation current is superimposed on the output current of the error amplifier in the voltage regulator at the power transistor control node.

[0114] In this embodiment, the charge / discharge array includes a pull-down compensation branch and a pull-up compensation branch. The pull-down compensation branch is used for load-heavy transients and consists of 32 parallel NMOS discharge transistors. Each discharge transistor is connected in series with a current-limiting resistor and then connected in parallel to the gate node of the power transistor. Each discharge transistor is controlled to turn on or off by an independent gating control bit, and the discharge current of a single discharge transistor is 100μA. The pull-up compensation branch is used for load-light transients and consists of 32 parallel PMOS charging transistors. Each charging transistor is connected in series with a current-limiting resistor and then connected in parallel to the gate node of the power transistor. Each charging transistor is controlled to turn on or off by an independent gating control bit, and the charging current of a single charging transistor is 100μA.

[0115] The mapped compensation intensity index is divided by 100, multiplied by the total number of cells in the corresponding branch, and rounded down to obtain the number of cells to be selected for the pull-down compensation branch and the pull-up compensation branch, respectively. The selection strategy uses a thermometer coding method, selecting cells sequentially starting from the first cell until the calculated number of cells is reached. For example, when the mapped compensation intensity index is 50%, 16 cells are selected for the pull-down compensation branch and 16 cells are selected for the pull-up compensation branch; when the mapped compensation intensity index is 100%, all 32 cells are selected.

[0116] The corresponding compensation branch is selected according to the compensation direction determined in step S2. If it is a rising edge load transition, the calculated number of discharge units in the pull-down compensation branch is selected, and a pull-down compensation current is injected into the gate of the power transistor. The total pull-down current is equal to the sum of the discharge currents of all selected discharge units, that is, the number of selected units multiplied by the discharge current of a single discharge transistor. This is used to accelerate gate discharge and quickly turn on the power transistor to supply the increased load current. If it is a falling edge load transition, the calculated number of charging units in the pull-up compensation branch is selected, and a pull-up compensation current is injected into the gate of the power transistor. The total pull-up current is equal to the sum of the charging currents of all selected charging units, that is, the number of selected units multiplied by the charging current of a single charging transistor. This is used to accelerate gate charging and quickly turn off the power transistor to suppress output voltage overshoot.

[0117] The pull-down or pull-up compensation current and the output current of the error amplifier in the voltage regulator are linearly superimposed at the gate node of the power transistor, without interference. During transients, the additional gate drive current provided by the compensation array is much greater than the static drive capability of the error amplifier, thereby enabling the power transistor gate voltage to change with an extremely high slew rate, achieving a fast response.

[0118] This embodiment achieves continuously adjustable compensation intensity through a thermometer code gating method, avoiding current glitches during the gating process. The pull-down compensation branch and pull-up compensation branch are controlled independently, providing precise compensation for rising and falling edge transients respectively. The compensation current and the error amplifier output current are linearly superimposed at the gate node, which does not affect the normal operation of the main loop and can provide transient compensation current far exceeding the static drive capability, effectively solving the contradiction between response speed and static power consumption in traditional solutions. At the same time, in steady state without transient events, the gating control bits of all charging and discharging units are zero, the current drawn by the compensation array from the power rail is zero, and only nanoampere-level leakage current exists, fundamentally achieving zero static power consumption.

[0119] Specifically, in step S4, recovery performance metrics for each load shift event are collected, including:

[0120] The actual recovery time of the load jump event is collected, which is the length of time from the moment when the amplitude of the rate of change first exceeds the preset threshold until the output voltage recovers to near the target value and continues to remain for more than the preset stable time.

[0121] The maximum voltage deviation of the load jump event is collected, and the absolute value of the maximum undershoot or overshoot of the output voltage during the recovery process is obtained.

[0122] In this embodiment, two recovery performance indicators are automatically collected after each load transition event. The first indicator is the actual recovery time, defined as the time from when the amplitude of the rate of change first exceeds a preset threshold until the output voltage recovers to near the target value and remains stable for more than a preset stabilization time. Specifically, the timer is triggered when the absolute value of the filtered rate of change first exceeds the preset threshold; the timer stops when the output voltage enters the range near the target value and remains stable for more than the preset stabilization time, and this time is recorded as the actual recovery time.

[0123] The second metric is the maximum voltage deviation, defined as the maximum magnitude by which the output voltage deviates from the target value during the recovery process. This includes the maximum undershoot occurring at the rising edge of the load transition and the maximum overshoot occurring at the falling edge of the load transition; the absolute value of these two values ​​is taken as the maximum voltage deviation. Specifically, during the entire recovery process after the load transition, the output voltage is continuously monitored using a high-speed comparator array. The maximum negative and maximum positive deviations of the output voltage relative to the target value are recorded, and the larger absolute value is taken as the maximum voltage deviation for this transient event. The acquisition of these metrics is achieved through an on-chip time-to-digital converter and a high-speed comparator array. The time-to-digital converter provides high-precision time measurement, and the high-speed comparator array captures the instantaneous peak deviation of the output voltage in real time.

[0124] Specifically, in step S4, adjusting the total number of cells in the charge-discharge array and the current value of a single charge-discharge cell according to the recovery effect index includes:

[0125] A comprehensive score is calculated based on the ratio of the actual recovery time to the target recovery time and the ratio of the maximum voltage deviation to the target voltage deviation. The calculation formula is as follows:

[0126]

[0127] Where Q represents the overall score. , The weighting coefficients and To preset the target recovery time, This refers to the actual recovery time. To prevent deviation from the preset target voltage, This represents the maximum voltage deviation.

[0128] When the overall score is less than the preset score, the total number of units in the charge-discharge array and the current value of a single charge-discharge unit are adjusted, and the adjusted parameters are stored in the on-chip non-volatile memory.

[0129] Specifically, adjusting the total number of cells in the charge / discharge array and the current value of a single charge / discharge cell includes:

[0130] When the comprehensive score of a series of transient events is less than the preset threshold and the compensation intensity index is in the saturation region, the total number of units in the charge-discharge array is increased.

[0131] When the comprehensive score of a series of transient events is less than the preset threshold and the compensation intensity index is in the low value range, the total number of units in the charge-discharge array is reduced.

[0132] When the negative impact of the actual recovery time on the overall score is greater than the maximum voltage deviation, the current value of a single charge / discharge unit is increased;

[0133] When the negative impact of the maximum voltage deviation on the overall score is greater than the actual recovery time, the current value of a single charge / discharge unit is reduced.

[0134] Each adjustment should not exceed a preset percentage of the original value.

[0135] In this embodiment, after each transient event, a comprehensive score is calculated based on the actual recovery time and maximum voltage deviation collected, which is used to quantitatively evaluate the compensation effect. The comprehensive score is calculated as follows: the preset target recovery time is divided by the actual recovery time to obtain the recovery time ratio, and the preset target voltage deviation is divided by the maximum voltage deviation to obtain the voltage deviation ratio. These two ratios are then assigned preset weighting coefficients and summed to obtain the comprehensive score. The sum of the weighting coefficients is 1. The shorter the actual recovery time and the smaller the maximum voltage deviation, the higher the comprehensive score, indicating a better compensation effect.

[0136] When the overall score falls below a preset threshold, parameter fine-tuning is triggered. The fine-tuning strategy is as follows: If the overall score of multiple consecutive transient events is below the preset threshold, and the compensation intensity index is frequently in the saturation region, it indicates that the maximum compensation capability of the current array is insufficient, so the total number of units in the charge / discharge array is increased; if the overall score of multiple consecutive transient events is below the preset threshold, and the compensation intensity index is frequently in the low value region, it indicates that the compensation resolution of the current array is insufficient, so the total number of units in the charge / discharge array is reduced to improve the control accuracy during low-intensity compensation; if the negative impact of the actual recovery time on the overall score is greater than that of the maximum voltage deviation, it indicates that the main problem is insufficient recovery speed, so the current value of a single charge / discharge unit is increased to enhance the compensation intensity; if the negative impact of the maximum voltage deviation on the overall score is greater than that of the actual recovery time, it indicates that the main problem is excessive voltage overshoot or undershoot, so the current value of a single charge / discharge unit is decreased to reduce the compensation intensity and avoid overcompensation.

[0137] Each adjustment is limited to a preset percentage of the original value to avoid performance oscillations caused by excessive adjustment in a single instance. The adjusted total number of cells and the current value of each individual cell are stored in on-chip non-volatile memory, ensuring that the data is not lost after power failure. This guarantees that the system can start from the previously optimized parameter state after each power-on, allowing the compensation parameters to continuously evolve with the long-term changes in load conditions.

[0138] This embodiment quantifies the transient recovery effect of each time through comprehensive scoring and automatically fine-tunes the compensation parameters accordingly, achieving closed-loop adaptive optimization. The fine-tuning strategy is differentiated for two dimensions: the total number of units and the unit current, based on the range of the compensation intensity index and the type of the deductible term, enabling the system to adapt to the optimal compensation requirements under different load conditions. The adjustment range is limited each time, ensuring the stability of the optimization process. The establishment of a historical optimal parameter library and a warm-up mechanism allow new transient events to directly call the matching optimal parameters without having to start from the default values ​​and gradually optimize, significantly improving the system's convergence speed and long-term adaptability.

[0139] Specifically, the method provided in this application also includes the establishment and updating of a historical optimal parameter database. A historical optimal parameter database is established according to the load switching direction and the amplitude range of the rate of change, storing the total number of optimal units, the optimal current value of a single unit, and the corresponding historical highest comprehensive score for each interval. The rate of change amplitude range is divided into three levels: low, medium, and high, based on the magnitude of the amplitude.

[0140] When a new load transition event occurs, a matching amplitude range entry is searched in the parameter library based on the transition direction and the magnitude of the rate of change. If a matching entry is found, the optimal total number of cells and the optimal single cell current value stored in that entry are used as the initial parameters for the charging and discharging array in the current transient event to achieve parameter preheating; if no matching entry is found, the system starts from the default parameters.

[0141] After each transient event, the transition direction, the range of the rate of change amplitude, the parameter combination used, and the overall score are recorded as a new entry. If a corresponding entry with the same transition direction and amplitude range already exists in the parameter library, the current overall score is compared with the highest historical overall score for that entry. If the current overall score is higher, the record for that entry is replaced with the current parameter combination, and the highest historical overall score is updated. If the current overall score remains below the preset failure threshold and reaches the required number of consecutive failures, it indicates that the optimal parameters for that load condition have become ineffective due to the long-term evolution of the load characteristics. In this case, the parameter library for that entry is rebuilt, all historical records for that entry are cleared, and optimization restarts from the default parameters. If a corresponding entry does not exist in the parameter library, the current parameters are stored as a new entry in the parameter library.

[0142] It should be noted that during the compensation execution in step S3, after a load transition is detected, the corresponding compensation branch is activated after a preset fixed delay. The compensation remains activated until either of the following conditions is met, at which point compensation terminates: Condition 1, the compensation timing reaches the compensation duration estimated in step S2; Condition 2, the amplitude of the rate of change falls below a preset threshold and remains below a preset number of sampling periods, indicating that the transient event has ended. Upon termination of compensation, a step-by-step shutdown strategy is adopted, gradually shutting down the selected charging and discharging units in multiple steps to avoid gate voltage rebound caused by a sudden complete shutdown, ensuring a smooth and stable shutdown process.

[0143] This embodiment establishes a historical optimal parameter library and a warm-up mechanism, enabling new transient events to directly call the matching optimal parameters without gradually optimizing from default values, significantly improving the system's convergence speed. Through the adaptive update and condition reconstruction mechanism of the parameter library, it ensures that the parameter library remains effective over the long term, following the evolution of load characteristics and avoiding the problem of historical parameters becoming invalid due to fundamental changes in load characteristics. A step-by-step shutdown strategy avoids gate voltage rebound caused by instantaneous complete shutdown, ensuring a smooth and stable compensation termination process. By turning off all charging and discharging units when there are no transients, static zero power consumption is achieved, fundamentally solving the problem of consuming large static current to maintain fast response capabilities in traditional solutions.

[0144] The above embodiments are merely illustrative examples and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this application.

Claims

1. A method for enhancing the transient response of a voltage regulator based on load change rate feedforward prediction, characterized in that, include: Step S1: Real-time acquisition of load current and determination of the rate of change of load current; based on the rate of change, determination of whether a load jump occurs and the direction of the load jump. Step S2: In response to a load change, determine the compensation intensity based on the ratio of the magnitude of the rate of change to the preset maximum rate of change, determine the compensation direction as pull-down compensation or pull-up compensation based on the direction of the load change, and determine the compensation duration based on the ratio of the magnitude of the rate of change to the expected current change. Step S3: Determine the number of charge and discharge units to be selected according to the compensation intensity index, select the corresponding number of charge and discharge units in the charge and discharge array composed of multiple charge and discharge units according to the compensation direction, output compensation current to the power tube control terminal of the voltage regulator, and control the compensation duration according to the compensation duration. Step S4: Collect the recovery effect index for each load transition event, adjust the total number of units in the charging and discharging array and the current value of a single charging and discharging unit according to the recovery effect index, and use the adjusted parameters for compensation control in subsequent load transition events.

2. The method according to claim 1, characterized in that, In step S1, determining the rate of change of the load current includes: The load current is continuously sampled at a fixed sampling frequency to obtain a digital sequence of the load current; The original value of the rate of change is calculated by dividing the difference between the digital load current at the current moment and the previous moment by the sampling interval. The calculation formula is as follows: in, The original value of the rate of change. This is the digital value of the load current at the current moment. This represents the digital value of the load current at the previous moment. The sampling interval; The original value of the rate of change is subjected to digital low-pass filtering to obtain the filtered rate of change, and the calculation formula is as follows: ; in, The filtered rate of change at the current time. These are the filter coefficients. This represents the rate of change after filtering at the previous moment.

3. The method according to claim 1, characterized in that, In step S1, determining whether a load jump has occurred and the direction of the load jump based on the rate of change includes: The rate of change is compared with preset positive and negative thresholds; If the rate of change is greater than the positive threshold, it is determined to be a rising edge load jump; If the rate of change is less than the negative threshold, it is determined to be a falling edge load jump; If the absolute value of the rate of change is less than the positive threshold, it is determined that no load jump has occurred.

4. The method according to claim 1, characterized in that, In step S2, the formula for calculating the compensation strength index is: Where P is the compensation strength, This is the preset maximum rate of change.

5. The method according to claim 1, characterized in that, In step S2, determining the compensation direction as pull-down compensation or pull-up compensation based on the load switching direction includes: If the load jumps on the rising edge, the compensation direction is pull-down compensation, which is used to pull down the control terminal of the power transistor for discharge. If the load transition occurs on a falling edge, the compensation direction is pull-up compensation, which is used to pull up and charge the control terminal of the power transistor.

6. The method according to claim 1, characterized in that, In step S2, the formula for calculating the compensation duration is: in, To compensate for the duration, To preset the maximum compensation time, This represents the expected change in current.

7. The method according to claim 1, characterized in that, In step S3, the charging and discharging array includes a pull-down compensation branch and a pull-up compensation branch. The pull-down compensation branch contains N parallel pull-down discharging units, and the pull-up compensation branch contains M parallel pull-up charging units. The number of gated units is determined based on the compensation intensity index, including: The number of gated units is determined based on the compensation intensity index, and the calculation formula is as follows: Where P is the compensation intensity index, N is the total number of units in the pull-down compensation branch, and M is the total number of units in the pull-up compensation branch. To determine the number of units selected for the pull-down compensation branch, The number of units selected for the pull-up compensation branch; If the load transition occurs on a rising edge, then the number of pull-down compensation branches selected is [number missing]. Each pull-down discharge unit injects pull-down compensation current into the control terminal of the power transistor, and the total pull-down current is: in, For the total pull-down current, This is the selection control bit for the i-th pull-down discharge unit. The discharge current of a single pull-down discharge unit; If the load transition occurs on a falling edge, then the number of pull-up compensation branches selected is [number missing]. Each pull-up charging unit injects pull-up compensation current into the control terminal of the power transistor, and the total pull-up current is: in, For the total pull-up current, This is the strobe control bit for the j-th pull-up charging unit. The charging current for a single pull-up charging unit; The pull-down compensation current or pull-up compensation current is superimposed on the output current of the error amplifier in the voltage regulator at the power transistor control node.

8. The method according to claim 1, characterized in that, In step S4, recovery performance metrics for each load shift event are collected, including: The actual recovery time of the load jump event is collected, which is the length of time from the moment when the amplitude of the rate of change first exceeds the preset threshold until the output voltage recovers to near the target value and continues to remain for more than the preset stable time. The maximum voltage deviation of the load jump event is collected, and the absolute value of the maximum undershoot or overshoot of the output voltage during the recovery process is obtained.

9. The method according to claim 1, characterized in that, In step S4, adjusting the total number of cells in the charge-discharge array and the current value of a single charge-discharge cell according to the recovery effect index includes: A comprehensive score is calculated based on the ratio of the actual recovery time to the target recovery time and the ratio of the maximum voltage deviation to the target voltage deviation. The calculation formula is as follows: Where Q represents the overall score. , The weighting coefficients and To preset the target recovery time, This refers to the actual recovery time. To prevent deviation from the preset target voltage, This represents the maximum voltage deviation. When the overall score is less than the preset score, the total number of units in the charge-discharge array and the current value of a single charge-discharge unit are adjusted, and the adjusted parameters are stored in the on-chip non-volatile memory.

10. The method according to claim 9, characterized in that, Adjusting the total number of cells in the charge / discharge array and the current value of a single charge / discharge cell includes: When the comprehensive score of a series of transient events is less than the preset threshold and the compensation intensity index is in the saturation region, the total number of units in the charge-discharge array is increased. When the comprehensive score of a series of transient events is less than the preset threshold and the compensation intensity index is in the low value range, the total number of units in the charge-discharge array is reduced. When the negative impact of the actual recovery time on the overall score is greater than the maximum voltage deviation, the current value of a single charge / discharge unit is increased; When the negative impact of the maximum voltage deviation on the overall score is greater than the actual recovery time, the current value of a single charge / discharge unit is reduced. Each adjustment should not exceed a preset percentage of the original value.