A carrier synchronization method and system for cascaded H-bridge power units based on single-fiber multiplexing
By employing single-fiber multiplexing technology and timer hardware acquisition method, the synchronization problem of cascaded H-bridge converters is solved, achieving high-precision and fast carrier synchronization, which is suitable for high-voltage frequency conversion soft starters.
Patent Information
- Application Number
- CN202611251543.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-18
- Publication Date
- 2026-11-10
AI Technical Summary
Existing carrier synchronization methods for cascaded H-bridge converters suffer from problems such as excessive number of signal lines, high cost of optical fibers, insufficient synchronization accuracy, and slow setup speed, making it difficult to meet the synchronization requirements of high-voltage frequency conversion soft starters.
A carrier synchronization method based on cascaded H-bridge power units using single-fiber multiplexing is adopted. By using synchronization flag bits in data frames and timer hardware capture technology, precise synchronization of carrier phase is achieved, reducing the number of optical fibers and improving synchronization accuracy, while taking into account both rapid setup and continuous operation.
The number of optical fibers was reduced, synchronization accuracy was improved, hardware modification costs were reduced, power unit interchangeability was supported, and rapid synchronization and stable operation were achieved, meeting the synchronization requirements of high-voltage frequency converter soft starters.
Smart Images

Figure CN123074892A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power electronic converter control technology, specifically to a carrier synchronization method and system for cascaded H-bridge power units based on single-fiber multiplexing, applicable to medium and high voltage cascaded H-bridge converters, static var generators, and high voltage frequency conversion soft starters. Background Technology
[0002] Cascaded H-Bridge (CHB) converters achieve high-voltage output by connecting multiple low-voltage power units in series. They are characterized by modularity, easy expansion, and high-quality output waveforms, and are widely used in medium- and high-voltage variable frequency speed control, static var generators, and high-voltage variable frequency soft starters. In 13.8kV high-voltage variable frequency soft starters, typically 12 power units are cascaded per phase, for a total of 36 power units across three phases.
[0003] Cascaded H-bridge converters often employ carrier phase-shifted pulse width modulation (CPS-SPWM) as their control strategy. This method requires that the carriers of N power units within each phase maintain equal phase shifts, with the phase shift angle between adjacent units being 180° / N. For a 12-unit system, a phase shift angle between adjacent units of 15° corresponds to a time offset of 83.333 μs at a carrier frequency of 500 Hz.
[0004] The harmonic cancellation effect of CPS-SPWM relies on the strict equal spacing of the carrier phases of each power unit. When the synchronization time error exceeds 5μs, the equivalent phase error exceeds 0.9°, and the total harmonic distortion rate of the output voltage increases significantly. Therefore, achieving carrier synchronization of multiple power units is a key technical issue for the engineering application of CPS-SPWM.
[0005] Existing multi-power unit synchronization methods mainly include the following: (1) Centralized PWM generation scheme: The main controller directly generates the PWM signals of all power units and sends them to each driver board through independent signal lines. The number of signal lines in this scheme increases linearly with the number of units. A 36-unit system requires 144 signal lines. The high-voltage insulation design is complex and it is only suitable for small-scale systems.
[0006] (2) Independent Synchronization Pulse Line Scheme: Each power unit is equipped with two optical fibers: a data fiber and a dedicated synchronization fiber. Each unit resets its local timer at the edge of the synchronization pulse. This scheme can achieve synchronization accuracy at the nanosecond level, but a 36-unit system requires 72 optical fibers, resulting in high costs for optical modules and cabling.
[0007] (3) Communication frame header external interrupt synchronization scheme: The start edge of the data frame is used as a time reference, and the arrival time of the edge is captured by the general GPIO external interrupt (EXTI). This scheme does not increase the number of optical fibers, but the external interrupt response time is affected by factors such as the current code execution and interrupt priority preemption. The typical jitter is 2μs to 10μs, which is difficult to eliminate through software optimization.
[0008] (4) Purely distributed digital phase-locked loop scheme: Each unit converges slowly through the statistical error using a digital phase-locked loop, which has strong anti-interference ability, but the initial convergence time may be tens of seconds, which is difficult to meet the requirements of rapid start-up.
[0009] The main problems with the existing technologies mentioned above are: centralized schemes have too many signal lines; independent synchronization pulse line schemes have high fiber costs; external interrupt synchronization schemes for frame headers have insufficient accuracy; and purely distributed phase-locked loop schemes have slow setup speeds. Therefore, a synchronization method is needed that can reduce the number of fibers, improve synchronization accuracy, and also maintain a good setup speed. Summary of the Invention
[0010] The purpose of this invention is to provide a method and system for carrier synchronization of cascaded H-bridge power units based on single-fiber multiplexing, which reduces synchronization error without adding dedicated synchronization fiber, supports power unit interchangeability and dynamic configuration, and balances synchronization establishment speed and operational continuity.
[0011] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing includes the following steps: Step 1: The distribution controller sends data frames on N downlink optical fibers at the same time. A synchronization flag is set in the control command field of the data frame to indicate whether the current frame is a synchronized frame.
[0012] Step 2: On each power unit controller, the starting edge of the data frame is simultaneously connected to the external trigger pin and the input capture pin after being shaped by the shaping circuit. The count value corresponding to the arrival time of the starting edge is automatically latched through the timer hardware input capture channel.
[0013] Step 3: Calculate the target count offset based on the phase angle field in the data frame, and compare the input capture value with the target count offset to obtain the phase error.
[0014] Step 4: Select the alignment mode based on the PWM output enable state: Hard alignment is performed when the PWM main output is off and a data frame with a valid synchronization flag is received, and the PWM timer count start point is rebuilt; Soft alignment is performed when the PWM main output is enabled, and the PWM timer auto-reload value is limited and corrected every carrier cycle.
[0015] This application also provides a system for the above-described cascaded H-bridge power unit carrier synchronization method based on single-fiber multiplexing: It includes a main controller, a distribution controller, N power unit controllers, a photoelectric conversion and shaping circuit, an input capture timer, a PWM timer, an auxiliary timer, and a processing unit, wherein: The main controller is used to calculate the modulation parameters of multiple power units; The distribution controller is connected to the main controller via a parallel bus and is used to encapsulate modulation parameters into data frames and transmit the data frames at the same time on multiple downlink optical fibers. The data frame includes a frame header, a control command field, a phase angle field, a carrier frequency field, a duty cycle field, and a check field. The control command field includes a synchronization flag bit. N power unit controllers are connected to the distribution controller via N downlink optical fibers, and each power unit controller includes: The photoelectric conversion and shaping circuit is used to receive downlink optical fiber signals and output level-shaped digital signals. The input capture channel of the input capture timer is connected to the output of the shaping circuit to hardware latch the count value when the start edge of the data frame arrives; The PWM timer operates in center-aligned mode and acts as the master timer, outputting a trigger synchronization signal to the input capture timer. The auxiliary timer operates in single-pulse mode and is used for delayed start-up during the hard alignment phase; The processing unit is used to calculate the target count offset based on the phase angle field, compare the input capture value with the target count offset to obtain the phase error, and select hard alignment or soft alignment mode according to the PWM main output enable state.
[0016] As a further improvement to this system, the three pins of the power unit controller are electrically shorted and connected to the output of the shaping circuit. The three pins are the external trigger pin, the first timer input capture pin, and the second timer input capture pin of the corresponding power unit controller. The data frame transmission rate is 2 Mbit / s to 3 Mbit / s, and the communication format is 8N1. N is an integer between 3 and 100.
[0017] Compared with the prior art, the present invention has the following beneficial effects: (1) No additional dedicated synchronization fiber is added. The synchronization reference signal and control data are multiplexed into the same fiber optic link, and the start edge of the data frame is used as a common time reference. Compared with the independent synchronization pulse line scheme, the number of fibers in the 36-unit system is reduced from 72 to 36, a reduction of 50%.
[0018] (2) Improve synchronization accuracy. The arrival time of the reference edge is automatically latched by the timer hardware input capture channel, avoiding time jitter caused by external interrupt response delay. The synchronization error is reduced from 2μs to 10μs in the external interrupt scheme to less than 3μs.
[0019] (3) Balancing establishment speed and operational continuity. Hard alignment is used to quickly establish the counting start point during the shutdown phase, while soft alignment is used to suppress long-term drift during the operation phase. The two are mutually exclusive in their application conditions and cover all operating conditions.
[0020] (4) Supports interchangeable power units. All power units are programmed with the same program, do not have a communication address, and their identity is determined by the physical fiber optic port, supporting interchangeability at any location.
[0021] (5) Low hardware modification cost. Only the three pins of the power unit controller need to be electrically shorted to the same shaping circuit output terminal. No additional components or external interfaces are required. It can be achieved through PCB jumper wires. Attached Figure Description
[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1 This is a schematic diagram of a single-phase N-unit cascaded H-bridge topology.
[0024] Figure 2 The ideal 12-unit CPS-SPWM carrier distribution and cascaded output waveform diagram.
[0025] Figure 3 This is the overall architecture diagram of the single-fiber multiplexing synchronization scheme of the present invention.
[0026] Figure 4 This is a timing diagram showing the relationship between data frames and synchronization references.
[0027] Figure 5 This is a graph showing the decomposition and contribution of error sources.
[0028] Figure 6 This is a comparison image before and after the renovation.
[0029] Figure 7 This is a comparative chart of the various methods.
[0030] Figure 8 This is a comparison chart of worst-case and root mean square (RMS) superposition.
[0031] Figure 9 This is a flowchart for hard-aligned execution.
[0032] Figure 10 This is the state transition diagram for a synchronous state machine.
[0033] Figure 11 The complete cycle cascaded output waveform is one fundamental frequency period (20ms).
[0034] Figure 12 The waveform diagram is for the output of a single-phase 12-unit cascade.
[0035] Figure 13 This is a spectrum diagram of the cascaded output voltage.
[0036] Figure 14 This is a timing diagram for out-of-step detection.
[0037] Figure 15 This is a long-term phase drift trend chart. Detailed Implementation
[0038] The basic principle of this method is: like Figure 2 The 12 triangular carriers shown in (a) are staggered by 15° in phase distribution. Each unit carrier is evenly distributed within one cycle and is compared with a sinusoidal modulation wave to generate a PWM pulse. Figure 2 (b) shows the three-level voltage waveforms (+Vdc, 0, -Vdc) output by each unit. Each unit operates independently at a switching frequency of 500Hz, and different switching times are achieved by carrier phase shifting. Figure 2 (c) presents a 25-level stepped wave output after 12 units are connected in series. Due to the phase shift of the carrier, the switching actions of adjacent units are staggered, the equivalent switching frequency is increased to 12kHz, and the output waveform is close to a sine wave.
[0039] like Figure 3 The diagram shows the overall architecture of a single-fiber multiplexing synchronization scheme, consisting of three layers: a DSP main control layer, an FPGA distribution layer, and an STM32 power unit layer. The DSP main control layer (TMS320F28335) is responsible for the calculation of modulation parameters and V / F control of 36 power units, and exchanges data with the FPGA distribution layer through the XINTF parallel bus; The FPGA distribution layer (XC7A35T) is responsible for data frame assembly, CRC check calculation, unified time transmission of 36 optical fibers, and CMD_SYNC synchronization flag management. The STM32 power unit layer (36 STM32F105RBT6) captures the arrival time of the measurement reference edge through the TIM3 hardware input, generates center-aligned PWM through TIM1, implements hard-aligned delay start through TIM2, and maintains carrier phase consistency through a synchronous state machine and a soft alignment mechanism.
[0040] The core innovation of this architecture lies in multiplexing the synchronization reference signal and control data into the same physical fiber link. The 36 downlink fibers simultaneously undertake the functions of control data transmission and synchronization time base distribution, achieving carrier phase shift synchronization of 36 power units without adding a dedicated synchronization link.
[0041] like Figure 4 The data frame and synchronization reference relationship shown: like Figure 4 As shown in (a), the FPGA generates a unified transmission trigger signal at a fixed phase of the carrier period, which serves as the starting point for the simultaneous transmission of data frames through 36 optical fibers.
[0042] Figure 4 (b) indicates that the 36 downlink optical fibers transmit the first byte 0xAA at the same time, and the starting edges of each channel remain aligned without any phase shift during transmission.
[0043] Figure 4 (c) shows that the start bit of the UART frame transitions from an idle high level to a low level, and this falling edge is a time marker that can be recognized by all units.
[0044] Figure 4 (d) shows that the signal after photoelectric conversion is shaped and inverted by a 74HC14 Schmitt trigger, and the falling edge of the starting bit is converted into a rising edge, thus improving the edge steepness.
[0045] Figure 4 (e) shows that the rising edge after shaping triggers TIM3 input capture, and the hardware automatically latches the count value into the CCR1 register without going through the CPU interrupt path.
[0046] Figure 5 The results indicate that among the eight error sources, the soft alignment dead zone (1000ns) and the individual differences of optical modules (500ns) are the main contributors, covering four types of error: random, systematic, quantization, and algorithmic.
[0047] Figure 6 The description states that after the EXTI scheme was modified to TIM3 hardware input capture, the four indicators of critical error, total error, phase error and 15° ratio all decreased by about 57%, and the cost of modifying the hardware was only shorting three pins.
[0048] Figure 7 This solution (solid green line) demonstrates a relatively balanced distribution across five dimensions: accuracy, fiber optic cost-effectiveness, unit interchangeability, hardware cost-effectiveness, and setup speed. It represents a compromise solution for various engineering indicators.
[0049] Figure 8The result indicates that the worst-case linear summation of the eight errors is 2.56 μs, and the root mean square summation is 1.28 μs, both of which are lower than the design target of 5 μs. The actual measured result is closer to the root mean square summation.
[0050] like Figure 9 As shown, the execution flow is as follows: Upon receiving CMD_SYNC=1, check the above conditions, stop TIM1 counting (CR1.CEN=0), clear TIM1_CNT, load the initial values of ARR and CCR, and calculate the delay count. When the delay count is zero, immediately generate an update event (UG) and start TIM1; when the delay count is greater than zero, configure TIM2 single-pulse mode for delayed start, and start TIM1 in the TIM2 expiration interrupt. After alignment, the set synchronization state is SYNC_STATE_LOCKED.
[0051] like Figure 10 The state transition relationships shown are as follows: INVALID transitions to WAITING after initialization; WAITING transitions to HARD_ALIGN when CMD_SYNC is received and MOE=0; HARD_ALIGN transitions to LOCKED after alignment is complete; LOCKED transitions to LOST when the out-of-synchronization criterion is met; LOST transitions to WAITING after reference recovery.
[0052] The output state of PWM is as follows: INVALID, WAITING, HARD_ALIGN, LOST: Block; LOCKED: Output is allowed (soft-aligned operation); Hard alignment requires MOE=0; soft alignment requires MOE=1.
[0053] Criterion for losing synchronization: (1) No capture for 5 consecutive cycles (10ms); (2) The phase difference exceeds ±486 counts (±5°) for 10 consecutive cycles; (3) The reference period deviates by more than ±10%.
[0054] Feasibility and effectiveness verification of this method: like Figure 11 As shown, a dual-channel oscilloscope was used to simultaneously observe the rising edge of the PWM in adjacent units (A1 and A2) of phase A (PA8 pin, TIM1_CH1 output), and the time difference between the two rising edges was measured. The theoretical value of the PWM waveform of the adjacent units is 83.333μs (corresponding to 15°).
[0055] Figure 12The measured results of the single-phase 12-unit cascaded output are presented from three perspectives: complete cycle waveform, local stepped magnification, and peak region ripple period. The complete cycle waveform shows a 25-level stepped waveform, consistent with the theoretical level number 2N+1=25, and the sinusoidal envelope is smooth. In the locally magnified image, the width of each level is basically uniform, and no obvious level loss, repetition, or jump phenomenon was observed. The peak region ripple period is approximately 83.3μs, corresponding to an equivalent switching frequency of 12kHz, which is consistent with the theoretical value feq=2Nfc. The test results show that the carrier phase relationship of each unit meets the expectations, and the harmonic cancellation and level superposition effect of CPS-SPWM meets the design requirements.
[0056] Figure 13 The FFT spectrum analysis results of the cascaded output voltage are presented. Figure 13 (a) Displays the complete spectrum from 0 to 20 kHz. The main energy is concentrated at two frequency points: the fundamental wave at 50 Hz and the main ripple wave at 12 kHz. The remaining frequency components are all below the -40 dB suppression reference line. Figure 13 (b) When the low-frequency band (02kHz) is amplified, the amplitudes of the 500Hz carrier sideband and the 1kHz carrier 2-fold sideband are both below -42dB, indicating that the CPS-SPWM effectively cancels the unit carrier frequency component. Figure 13 (c) Amplify the main ripple region (10~15kHz) and measure the main ripple frequency and... The results are consistent with the expected harmonic cancellation characteristics of CPS-SPWM, further verifying the correctness of the carrier phase shift synchronization of each unit.
[0057] The actual test results of the out-of-step detection test are as follows: Figure 14 As shown: Figure 14 (a) shows that at t=0ms, the downlink fiber of unit A6 is disconnected, the reference signal disappears and remains at a low level; Figure 14 (b) shows that no valid reference edge was captured in the subsequent 5 carrier cycles (Tc1~Tc5); Figure 14 (c) shows that no_reference_counter increments sequentially, reaching alarm threshold 3 at t=6ms and out-of-step threshold 5 at t=10ms; Figure 14 (d) indicates that the synchronous state machine transitions from SYNC_STATE_LOCKED to SYNC_STATE_LOST via the WARNING phase. Figure 14 (e) shows that the PWM output switches from enabled (MOE=1) to blocked (MOE=0) at t=10ms. At the same time, the A6 unit reports the out-of-synchronization information through the uplink frame status word Bit15 (ST_SYNC_LOST). After detection by the DSP, it is determined that the ready conditions are not met, and the whole machine enters the unready state. The response time from the fiber optic disconnection to the PWM blocking is controlled within 10ms.
[0058] After reconnecting the A6 fiber optic cable, the unit executes the recovery process: upon receiving a valid frame, it transitions from SYNC_STATE_LOST to SYNC_STATE_WAITING. Upon receiving CMD_SYNC=1, it performs hard alignment. After alignment is complete, it enters SYNC_STATE_LOCKED and reports ST_SYNC_LOCKED. The DSP detects that all units are synchronized and the readiness conditions are met. The measured recovery time is less than 2 seconds, including the complete process of hard alignment execution and status reporting, verifying the correctness of the out-of-synchronization detection and recovery mechanism.
[0059] like Figure 15 The following are the long-term stability test results of the phase deviation of cell A7 relative to cell A1: Figure 15 The (a) time-series trend chart shows that the phase deviation of the 48 sampling points (collected every 10 minutes) remained within a small range of ±0.5°, with a maximum value of +0.42°, a minimum value of -0.30°, and a mean of +0.06°. Figure 15 The distribution histogram (b) shows that the data is centrally distributed around the mean with a standard deviation of 0.21°, corresponding to a time error of approximately 1.17 μs. No systematic drift or error divergence was observed during the entire 8-hour run, indicating that the soft alignment mechanism can suppress the long-term error accumulation caused by factors such as temperature changes and crystal frequency drift, and the synchronization quality remains stable under continuous operation conditions.
[0060] The technical solution of the present invention will be clearly and completely described below with reference to specific embodiments. Obviously, the specific numerical values in the embodiments are for illustrative purposes only and do not constitute a limitation on the scope of protection of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0061] Example 1: 13.8kV High Voltage Variable Frequency Soft Starter Carrier Synchronization System 1. System Composition like Figure 3 As shown, the carrier synchronization system of this embodiment includes: (1) Main controller: The TMS320F28335 digital signal processor with a system clock of 150MHz is used to calculate the modulation parameters of 36 power units in three phases.
[0062] (2) Distribution controller: It adopts XC7A35T field programmable gate array (FPGA), with a working clock of 100MHz. It is connected to the main controller through an external parallel bus (XINTF) and is responsible for data frame assembly, CRC check calculation and multi-channel unified transmission.
[0063] (3) 36 power unit controllers: STM32F105RBT6 microcontrollers are used, with an external crystal oscillator of 25MHz and a system clock of 70MHz. They are connected to the distribution controllers via downlink optical fibers. There are 12 power units per phase, for a total of 36 power units across the three phases.
[0064] (4) Downlink fiber optic link: 36 plastic optical fibers are used, the transmission rate is 2.5 Mbit / s, and the communication format is 8N1 (1 start bit, 8 data bits, no parity bit, 1 stop bit).
[0065] (5) Photoelectric conversion and shaping circuit: including photodetector and 74HC14 Schmitt trigger, converting optical signal into 3.3V logic level digital signal.
[0066] 2. Data Frame Format The downlink data frame is a fixed-length frame of 10 bytes, and the meaning of each byte is shown in Table S1-1.
[0067] Table S1-1 Meaning of each byte in the downlink data frame
[0068] The control command byte is defined as follows: Bit0: CMD_RUN (Run command) Bit1: CMD_STOP (Stop command) Bit2: CMD_FAULT (Fault Indicator) Bit3: CMD_RESET (Reset command) Bit4: CMD_BYPASS_ON (Bypass closing) Bit 5: CMD_BYPASS_OFF (Bypass Trip) Bit6: CMD_DIRECTION (direction bit) Bit7: CMD_SYNC (Synchronization Alignment Command) When Bit7 is set to 1, it indicates that the current frame is a synchronization-aligned frame, and the power unit performs hard alignment when the hard alignment conditions are met.
[0069] 3. Hardware electrical connections The three pins of the STM32F105RBT6 power unit controller—PA6 (TIM3_CH1 input capture pin), PA7 (TIM3_CH2 input capture pin), and PA12 (TIM1_ETR external trigger pin)—are electrically shorted via PCB jumpers or board modifications, and simultaneously connected to the output of the 74HC14 shaping circuit. The falling edge of the data frame start bit is inverted and shaped by the 74HC14, converted into a rising edge, and simultaneously input to the three pins.
[0070] 4. Timer Configuration (1) PWM timer TIM1 configuration: PWM timer TIM1 operates in center-aligned mode 3. The counter increments from 0 to the auto-reload value ARR = 35000 at a clock frequency of 35 MHz, then decrements back to 0, thus forming a complete PWM cycle. The prescaler PSC is set to 1, meaning the timer input clock is divided by 2 to obtain a 35 MHz counting clock. In center-aligned mode, the PWM cycle is:
[0071] The corresponding carrier frequency is 500Hz. TIM1 is also configured as the master timer, and the update event is set to the TRGO trigger signal output so as to trigger the ADC, other timers or related peripherals at the update time of each PWM cycle, so as to realize the synchronous operation between the modules.
[0072] (2) Input capture timer TIM3 configuration: The input capture timer TIM3 operates in slave mode with a reset function. Its trigger source is connected to the TRGO signal output from TIM1. When TIM1 generates an update trigger event, the TIM3 counter is automatically reset to zero, thus synchronizing the input capture timing with the PWM period of TIM1. The prescaler coefficient PSC of TIM3 is set to 3, meaning the timer input clock is divided by 4 to obtain a 17.5MHz counting clock. Therefore, the time resolution for each count is:
[0073] The automatic reload value ARR is set to 65535, providing a sufficient counting range; within the 2 ms PWM cycle of TIM1, TIM3 counts a total of:
[0074] The input capture channel TI1 of TIM3 is mapped to the PA6 pin and configured to be triggered by a rising edge. When a valid rising edge is detected, the current count value is latched to determine the time position of the input signal relative to the start of the PWM cycle. At the same time, the digital filter parameter IC1F is set to 0011 to filter out short-time glitches and high-frequency interference at the input end, thereby improving the stability and reliability of the capture results.
[0075] (3) Configuration of auxiliary timer TIM2: The auxiliary timer TIM2 is primarily used for delayed startup during the hard alignment phase to ensure that relevant control actions are executed accurately after the set delay time. TIM2 is configured in single-pulse mode (OPM) and, after startup, counts at a 17.5 MHz clock speed, with each count corresponding to approximately... The time resolution is such that when the counter reaches the preset value and generates an update event, TIM2 automatically stops counting, thus performing only one timing delay. By setting the prescaler value and auto-reload value of TIM2, the delay time required for the hard alignment stage can be flexibly configured, and an interrupt or output event can be triggered at the end of the delay, so that subsequent control operations can start in a predetermined sequence.
[0076] 5. Phase Measurement and Error Calculation The target count offset for each power unit is calculated based on the phase angle field:
[0077] For this embodiment, =17.5MHz, =500Hz, substituting and simplifying, we get:
[0078] The target count offsets for the 12 power units are shown in Table S1-2.
[0079] Table S1-2 Target Count Offset of 12 Power Units
[0080] The formula for calculating phase error is: ,in =35000, the Wrap function normalizes the error to the range of (-17500, +17500].
[0081] 6. Hard-aligned execution Hard alignment execution conditions: MOE=0 (PWM main output off) RUN=0 (No command to run) No local fault latch Received a valid frame with CMD_SYNC=1 The phase angle parameter phase_x100 is in the range of 0 to 35999. The carrier frequency fc_hz is in the range of 100 to 2000 Hz. Hard alignment execution process as follows Figure 9 As shown: (a) Check whether the above conditions are met; (b) Stop TIM1 counting (clear CR1.CEN bit); (c) Clear TIM1_CNT; (d) Load the initial values for ARR and CCR; (e) Calculate the delay count:
[0082] (f) When At this time: an update event is generated (TIM1_EGR.UG=1) and TIM1 is started, setting the synchronization state to SYNC_STATE_LOCKED; (g) When At this time: Configure TIM2 to single-pulse mode, and Load into TIM2_ARR, start TIM2, and set the synchronization status to SYNC_STATE_HARD_ALIGN; (h) When TIM2 expires, an interrupt is generated. In the interrupt service routine, TIM1_CNT is cleared, an update event is generated, TIM1 is started, and the synchronization state is set to SYNC_STATE_LOCKED.
[0083] 7. Soft alignment execution Soft alignment uses a limiting ratio control law:
[0084] In this embodiment =1, meaning that at most one TIM1 count can be corrected per carrier cycle. The TIM1 counter clock is 35MHz, and the time corresponding to a single correction is 28.57ns.
[0085] The dead zone threshold is set to 35 counts (approximately 1.0 μs), and no correction is performed when the absolute value of the phase error is less than the dead zone threshold.
[0086] 8. Synchronization State Machines and Out-of-Step Criteria State transition relations are as follows Figure 10 As shown.
[0087] The out-of-synchronization criterion is set as follows: (a) Reference loss alarm: If no capture occurs for 3 consecutive carrier cycles (6ms), the synchronization alarm bit is set; (b) Reference loss of synchronization: If no capture occurs for 5 consecutive carrier cycles (10ms), synchronization is lost and PWM is blocked; (c) Phase error exceeds limit: If the error count exceeds ±486 (corresponding to ±5°) for 10 consecutive cycles, it is judged as out of step and PWM is blocked; (d) Reference cycle abnormality: if the adjacent capture interval deviates from the standard value by more than ±10%, the capture is deemed invalid.
[0088] 9. Experimental Data The test was conducted on a prototype of a 13.8kV, 36-unit cascaded H-bridge high-voltage variable frequency soft starter. The main parameters are: rated voltage 13.8kV, rated power 3000kW, 12 units per phase, rated unit voltage 1100V, carrier frequency 500Hz, and carrier phase shift angle 15°.
[0089] (1) The phase difference test results of adjacent units are shown in Table S1-3.
[0090] Table S1-3 Test Results of Phase Difference Between Adjacent Units
[0091] Statistical results: The maximum absolute deviation was 0.36°, the root mean square deviation was 0.22°, and the proportion of the 15° phase shift angle was 2.4%.
[0092] (2) Cascaded output waveform test: The single-phase 12-unit cascaded output is a 25-level stepped waveform with a main ripple frequency of 12kHz, which is consistent with the theoretical value. A perfect match.
[0093] (3) Spectrum test: 50Hz fundamental frequency is 0dB, 500Hz carrier sideband is -45dB, 1kHz carrier double frequency sideband is -42dB, 12kHz main ripple is -28dB.
[0094] (4) Loss of synchronization and recovery test: After the fiber optic cable of unit A6 is disconnected, it enters the SYNC_STATE_LOST state and blocks PWM within 10ms; after the fiber optic cable is reconnected, the recovery time is less than 2s.
[0095] (5) Long-term stability test: After running continuously for 8 hours, the phase deviation of each unit A1~A12 remained within ±0.5°, with a maximum value of +0.42°, a minimum value of -0.30°, a mean of +0.06°, and a standard deviation of 0.21°.
[0096] Example 2: Application of Cascaded H-Bridge Converters of Different Sizes This embodiment illustrates the configuration of the synchronization method in cascaded H-bridge converters with different numbers of power units. The number of power units connected in series per phase of the cascaded H-bridge converter is denoted as... Each power unit uses a triangular carrier of the same frequency, and the carriers of adjacent power units are set with a fixed phase difference to form carrier phase shift modulation.
[0097] For each phase containing For a system with one power unit, the carrier phase shift angle between adjacent power units is determined by the following formula:
[0098] In the formula, The carrier phase shift angle between adjacent power units, in degrees (°); The number of power units connected in series in each phase is dimensionless.
[0099] No. The carrier phase angle of each power unit can be configured according to the following formula:
[0100] In the formula, For the first The carrier phase angle of each power unit, in degrees (°). This is the sequential numbering of the power unit within the cascaded unit of this phase. When the first power unit is used as the phase reference, its number is 0, corresponding to a carrier phase angle of 0°.
[0101] The carrier frequency is denoted as The corresponding carrier period is:
[0102] In the formula, The carrier frequency is measured in Hertz (Hz). The carrier period is measured in seconds (s).
[0103] In this embodiment, the carrier frequency is 500Hz, therefore:
[0104] When the carrier phase is converted to 360° corresponding to one complete carrier cycle, the time offset between adjacent power units is:
[0105] Will Substituting, we get:
[0106] In the formula, The time offset between the carrier start positions of adjacent power units is expressed in seconds (s).
[0107] With a carrier frequency of 500Hz, the carrier configurations corresponding to different numbers of power units are shown in Table S2-1.
[0108] Table S2-1 Carrier Configurations for Different Numbers of Power Units
[0109] Taking a 6kV system with 5 power units per phase as an example, the carrier phase shift angle between adjacent power units is:
[0110] The time offset between adjacent power units is:
[0111] The carrier phase angles corresponding to the five power units can be set to 0°, 36°, 72°, 108° and 144° respectively.
[0112] Taking a 10kV system with 8 power units per phase as an example, the carrier phase shift angle between adjacent power units is:
[0113] The time offset between adjacent power units is:
[0114] The carrier phase angles corresponding to the eight power units can be set to 0°, 22.5°, 45°, 67.5°, 90°, 112.5°, 135° and 157.5° respectively.
[0115] Taking a 13.8kV system with 12 power units per phase as an example, the carrier phase shift angle between adjacent power units is:
[0116] The time offset between adjacent power units is:
[0117] The carrier phase angles of the 12 power units are set sequentially at 15° intervals, with a phase angle configuration range of 0° to 165°.
[0118] Each power unit uses the same control program. The carrier phase angle of each power unit is not fixed across different program versions, but is configured by the main controller via downlink data frames. The downlink data frame sets the phase angle field `phase_x100`, which uses 0.01° increments. The relationship between its value and the carrier phase angle is as follows:
[0119] or:
[0120] In the formula, phase_x100 is the phase angle configuration value in the downlink data frame, which is dimensionless; The target carrier phase angle for the corresponding power unit is expressed in degrees (°). This indicates taking the nearest integer.
[0121] Examples of partial phase angle field configurations for systems of different sizes are shown in Table S2-2.
[0122] Table S2-2 Examples of partial phase angle field configurations for systems of different scales
[0123] When the system voltage level or the number of power units in operation per phase changes, the main controller adjusts the response based on the actual number of units in operation. Recalculate and corresponding to each power unit The corresponding phase_x100 is sent to each power unit via downlink data frames. After receiving and verifying the downlink data frames, the power unit converts the phase angle field into the count offset of its local timer, and starts or adjusts the PWM carrier according to the configured phase angle when the synchronization reference is reached. Thus, without changing the power unit control program, the corresponding carrier phase angle can be configured according to the actual number of power units in each phase.
[0124] Example 3: Parameter Configuration at Different Carrier Frequencies This embodiment illustrates the configuration method of PWM timer parameters and synchronization parameters when the carrier frequency changes. Each power unit's TIM1 uses center alignment mode 3, and the TIM1 counting clock remains at 35 MHz. The carrier frequency is transmitted from the fc_hz field in the downlink data frame, and each power unit calculates and updates the automatic reload value ARR of TIM1 based on the received carrier frequency.
[0125] In center-aligned counting mode, the TIM1 counter increments from 0 to ARR and then decrements from ARR back to 0, completing one carrier cycle. The count value corresponding to one carrier cycle is:
[0126] In the formula, This is the TIM1 count value corresponding to one PWM carrier cycle, in units of counts; This is the auto-reload value for TIM1, expressed in counts.
[0127] The carrier period and carrier frequency of TIM1 are as follows:
[0128]
[0129] In the formula, The carrier period is measured in seconds (s). The carrier frequency is measured in Hertz (Hz). This is the counting clock frequency of TIM1, in Hertz (Hz). This is the auto-reload value for TIM1, expressed in counts. Therefore, given a carrier frequency, the ARR value of TIM1 is determined by the following formula:
[0130] In this embodiment, the TIM1 counting clock is:
[0131] When the calculation result is not an integer, it needs to be taken as an integer value based on the timer's counting range and the allowable carrier frequency error. For example:
[0132] In the formula, The integer value actually written to the TIM1 auto-reload register, in counts; The target carrier frequency for the data frame is transmitted, and the unit is Hertz (Hz). This indicates taking the nearest integer.
[0133] Based on the actual ARR value written, the actual carrier frequency generated by TIM1 is:
[0134] In the formula, The actual carrier frequency is expressed in Hertz (Hz). The relative error between the target carrier frequency and the actual carrier frequency can be expressed as:
[0135] In the formula, The carrier frequency relative error is expressed as a percentage (%).
[0136] 1. Configuration when carrier frequency is 500Hz When the carrier frequency field in the data frame is:
[0137] The corresponding carrier period is:
[0138] The ARR value of TIM1 is:
[0139] Since TIM1 uses a center-aligned counting mode, the count value for one carrier cycle is:
[0140] Therefore, when the carrier frequency is 500Hz, the auto-reload value of TIM1 is set to 35000, and one PWM carrier cycle corresponds to 70000 TIM1 counts.
[0141] 2. Configuration when carrier frequency is 1000Hz When the carrier frequency field in the data frame is:
[0142] The corresponding carrier period is:
[0143] The ARR value of TIM1 is:
[0144] The count value corresponding to one carrier cycle is:
[0145] Therefore, when the carrier frequency is 1000Hz, the auto-reload value of TIM1 is set to 17500, and one PWM carrier cycle corresponds to 35000 TIM1 counts.
[0146] 3. Configuration at a carrier frequency of 2000Hz When the carrier frequency field in the data frame is:
[0147] The corresponding carrier period is:
[0148] The ARR value of TIM1 is:
[0149] The count value corresponding to one carrier cycle is:
[0150] Therefore, when the carrier frequency is 2000Hz, the auto-reload value of TIM1 is set to 8750, and one PWM carrier cycle corresponds to 17500 TIM1 counts.
[0151] The TIM1 parameters corresponding to different carrier frequencies are shown in Table S3-1.
[0152] Table S3-1 TIM1 parameters for different carrier frequencies
[0153] 4. Input capture timer cycle count conversion In the aforementioned embodiment, TIM3 is used to measure the time position of the synchronization signal relative to the start of the PWM cycle. The counting clock of TIM3 is 17.5MHz, and the time corresponding to a single count is:
[0154] In the formula, The time corresponding to a single TIM3 count, in seconds (s) or nanoseconds (ns); The clock frequency for TIM3 is in Hertz (Hz).
[0155] The TIM3 count value corresponding to one carrier cycle is:
[0156] In the formula, The TIM3 count value corresponds to one carrier cycle, and the unit is count. With a TIM3 counting clock of 17.5MHz, the cycle counts corresponding to different carrier frequencies are shown in Table S3-2.
[0157] Table S3-2 Period Counts for Different Carrier Frequencies
[0158] The difference in period count values between TIM1 and TIM3 is due to the different counting clocks they use. TIM1 uses a 35MHz counting clock, while TIM3 uses a 17.5MHz counting clock. Therefore, within the same time period, the count value of TIM1 is twice that of TIM3.
[0159] 5. Conversion between carrier frequency and phase delay When the carrier frequency changes, the time offset corresponding to the same carrier phase angle also changes. Let the carrier phase angle of a certain power unit be... The corresponding time offset is:
[0160] In the formula, For the first The time offset corresponding to each power unit is expressed in seconds (s). For the first The carrier phase angle of each power unit, in degrees (°). The carrier period is measured in seconds (s). The carrier frequency is expressed in Hertz (Hz).
[0161] When TIM3 or TIM2 is used to represent this time offset, the corresponding timer count value is:
[0162] Substituting the time offset expression, we get:
[0163] In the formula, phase angle The corresponding timer count value, in units of counts; The timer clock used for phase measurement or delay control is measured in Hertz (Hz); in this embodiment, the counting clocks for both TIM2 and TIM3 can be 17.5MHz.
[0164] Taking a carrier phase angle of 36° as an example, the corresponding time offset and 17.5MHz timer count value under different carrier frequencies are shown in Table S3-3.
[0165] Table S3-3 Time offset and 17.5MHz timer count value at different carrier frequencies
[0166] Therefore, after the carrier frequency is updated, in addition to the ARR value of TIM1, the acquisition comparison value, delay count value, or decision threshold related to phase-time conversion should also be recalculated according to the current fc_hz field. The carrier phase angle itself can remain unchanged, but its corresponding time offset and timer count value need to be converted based on the new carrier period.
[0167] 6. The update process of the fc_hz field and its parameters The downlink data frame sets the carrier frequency field fc_hz, whose value represents the target carrier frequency in Hertz. For example, fc_hz equals 500, 1000, and 2000, which respectively represent target carrier frequencies of 500Hz, 1000Hz, and 2000Hz.
[0168] After receiving the downlink data frame, the power unit updates the carrier frequency parameters according to the following procedure: (1) Check the frame format and check fields of the data frame and read fc_hz; (2) Determine whether fc_hz is within the allowed frequency configuration range of TIM1, and determine whether the calculated ARR value is within the valid range of the timer register; calculate the ARR value of TIM1 according to the following formula:
[0169] (3) Calculate the actual carrier frequency based on the actual ARR value, and update the input acquisition period value, phase offset count value and delay count value related to the carrier period; (4) Write the new ARR value into the preload register of TIM1; (5) Make the new parameters effective at the agreed synchronization time or TIM1 update event.
[0170] When using the preloading method, the new value of ARR is first written to the preloading register and then transferred to the actual effective register when an update event occurs. Each power unit can perform parameter switching based on the same synchronization reference, avoiding the transitional state where some power units have adopted the new carrier cycle while the rest of the power units are still using the original carrier cycle.
[0171] For ease of explanation, the data frame fields and local configuration results for the three carrier frequencies are shown in Table S3-4.
[0172] Table S3-4 Data Frame Fields and Local Configuration Results for Three Carrier Frequencies
[0173] Each power unit is programmed with the same control program, eliminating the need for separate program versions for 500Hz, 1000Hz, and 2000Hz. The main controller sends the current carrier frequency via the fc_hz field, and each power unit updates its local PWM timer and related synchronization counting parameters according to the same calculation relationship. If the carrier frequency is outside the preset range, or if the ARR value calculated based on that frequency exceeds the effective counting range of TIM1, the power unit will not perform the parameter update and will retain its original configuration or enter a pre-set parameter exception handling process.
[0174] The above embodiments are typical applications of the present invention and do not constitute a limitation on the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing, characterized in that, The method is applied to a cascaded H-bridge converter system including a main controller, a distribution controller, and N power unit controllers. The main controller and the distribution controller are connected via a parallel bus, and the distribution controller and the N power unit controllers are respectively connected via N downlink optical fibers. The method includes: Step 1: The distribution controller sends data frames on N downlink optical fibers at a unified time. The control command field of the data frame is set with a synchronization flag bit to indicate whether the current frame is a synchronized frame. Step 2: On each power unit controller, the starting edge of the data frame is shaped by the circuit and simultaneously connected to the external trigger pin and input capture pin of the controller. The count value corresponding to the arrival time of the starting edge is automatically latched through the timer hardware input capture channel. Step 3: Calculate the target count offset based on the phase angle field in the data frame, and compare the input capture value with the target count offset to obtain the phase error; Step 4: Select one of the following two alignment modes based on the PWM output enable state of the power unit: Step 4-1: When the PWM main output is turned off and a data frame with a valid synchronization flag is received, perform hard alignment and rebuild the PWM timer count start point; Step 4-2: When the PWM main output is enabled, perform soft alignment and limit the PWM timer auto-reload value to no more than the preset maximum correction amount in each carrier cycle.
2. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 1, characterized in that, The power unit controller includes a PWM timer and an input capture timer. The PWM timer operates in center-aligned mode and serves as the master timer to output a trigger signal. The input capture timer operates in slave mode reset state, and its counter is reset by the update event of the PWM timer, so that the count value of the input capture timer is synchronized with the PWM timer.
3. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 1, characterized in that, The hard alignment process includes: Step 4-1-1: Check whether the current state of the power unit meets the hard alignment conditions. The conditions include that the PWM main output enable bit is zero, there is no running command, there is no local fault latch, and the phase angle parameter and carrier frequency parameter are within the legal range. Step 4-1-2: Stop the PWM timer counting and clear its count register, then load the auto-reload value and the initial comparison value; Step 4-1-3: Calculate the delay count value based on the phase angle field; when the delay count value is zero, immediately generate an update event and start the PWM timer; when the delay count value is greater than zero, configure the auxiliary timer to single-pulse mode, load the delay count value into the auxiliary timer's auto-reload register, and start the auxiliary timer. Step 4-1-4: When the auxiliary timer expires and generates an interrupt, start the PWM timer in the interrupt service routine and set the synchronization state to locked.
4. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 1, characterized in that, The correction law for the soft alignment is a limiting ratio control: In the formula: This is the correction amount for the kth time. This is the phase error count value. This is the proportionality coefficient. The maximum correction amount in a single operation; The value is set to 1 PWM timer count; and no correction is performed when the absolute value of the phase error is less than the preset dead zone threshold.
5. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 1, characterized in that, The conversion formula for the target count offset is: In the formula: For target count offset, This is the value of the phase angle field. Input capture timer count clock frequency, For carrier frequency; The formula for calculating the phase error is as follows: In the formula: For input capture value, The Wrap function normalizes the error to the count value corresponding to one carrier cycle. Within the range.
6. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 1, characterized in that, Also includes: Step 5: Synchronization state management, which divides the synchronization state into five types: invalid state, waiting state, hard-aligned execution state, locked state, and out-of-sync state; The state transition condition is: an invalid state transitions to a waiting state after initialization is complete; The waiting state transitions to the hard-aligned execution state when a data frame with a valid synchronization flag is received and the PWM main output is turned off. The hard alignment execution state transitions to the locked state after alignment is complete; the locked state transitions to the out-of-step state when the out-of-step criterion is met. The out-of-synchronization state transitions to a waiting state after the reference signal is recovered.
7. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 6, characterized in that, The out-of-synchrony criteria include: A: Set an alarm threshold M1. When the reference edge is not captured for M1 consecutive carrier cycles, a synchronization alarm is triggered. B: Set a step loss threshold M2 and M2>M1. When the reference edge is not captured for M2 consecutive carrier cycles, step loss is triggered and the PWM output is blocked. C: Set the blocking threshold M3. When the absolute value of the phase error exceeds the preset phase error limit for M3 carrier cycles, it will trigger a loss of synchronization and block the PWM output. D: Set a percentage threshold. When the interval between two adjacent input capture values deviates from the standard carrier cycle count value by more than the percentage threshold, the capture is deemed invalid and will not be included in the phase error calculation.
8. The carrier synchronization method for cascaded H-bridge power units based on single-fiber multiplexing according to claim 1, characterized in that, The three pins of the power unit controller are electrically shorted to the same shaping circuit output terminal. The three pins are the external trigger pin and two timer input capture pins of the corresponding controller.
9. A system for carrier synchronization of cascaded H-bridge power units based on single-fiber multiplexing as described in any one of claims 1-8, characterized in that, include: The main controller is used to calculate the modulation parameters of multiple power units; The distribution controller, connected to the main controller via a parallel bus, is used to encapsulate modulation parameters into data frames and transmit the data frames at a unified time on multiple downlink optical fibers. The data frames include a frame header, a control command field, a phase angle field, a carrier frequency field, a duty cycle field, and a check field. The control command field includes a synchronization flag bit. N power unit controllers are each connected to the distribution controller via N downlink optical fibers. Each power unit controller includes: The photoelectric conversion and shaping circuit is used to receive downlink optical fiber signals and output level-shaped digital signals. An input capture timer, whose input capture channel is connected to the output of a shaping circuit, is used to hardware latch a count value when the start edge of a data frame arrives. The PWM timer operates in center-aligned mode and acts as the master timer, outputting a trigger synchronization signal to the input capture timer. An auxiliary timer, operating in single-pulse mode, is used for delayed start-up during the hard alignment phase; The processing unit is used to calculate the target count offset based on the phase angle field, compare the input capture value with the target count offset to obtain the phase error, and select hard alignment or soft alignment mode according to the PWM main output enable state.
10. A cascaded H-bridge power unit carrier synchronization system based on single-fiber multiplexing according to claim 9, characterized in that, The three pins of the power unit controller are electrically shorted and connected to the output of the shaping circuit. The three pins are the external trigger pin, the first timer input capture pin, and the second timer input capture pin of the corresponding power unit controller. The data frame transmission rate is 2 Mbit / s to 3 Mbit / s, and the communication format is 8N1. N is an integer between 3 and 100.
