With the increasing penetration of distributed photovoltaic systems into the power grid, the operational safety, stability, and economic efficiency of the grid face significant challenges. As the solar inverter market matures, many countries are phasing out policy support, leading asset operators to focus on optimizing the operation and maintenance of solar power generation systems. Maximizing power output, predicting and diagnosing faults, and achieving comprehensive observability, measurability, and controllability of the entire system have become critical goals. The rapid development of third-generation semiconductor technology has facilitated the widespread adoption of silicon carbide (SiC) devices in solar inverters. In practical designs, parallel connection of SiC devices is often necessary to enhance power handling capacity. However, current imbalance among parallel SiC devices can arise due to factors such as drive signal desynchronization, parameter tolerances, and asymmetrical board layout. This imbalance leads to excessive heating in devices carrying higher currents, adversely affecting their lifespan and the overall reliability of the solar inverter. Therefore, monitoring the current distribution in parallel SiC devices is essential for improving system reliability.
Traditional monitoring approaches involve placing current sensors in the current path of each SiC device, comparing sampled results to detect imbalances. While direct, this contact-based method introduces additional parasitic inductance into high-frequency switching circuits, potentially compromising the electrical performance and reliability of the solar inverter. Moreover, it requires multiple sensors, each rated for the maximum current of individual SiC devices, resulting in higher costs. Alternatively, non-contact magnetic sensors, such as anisotropic magnetoresistive (AMR) sensors, offer high bandwidth and avoid introducing parasitic inductance. However, this approach still necessitates multiple sensors and careful consideration of their placement on the printed circuit board (PCB). To address these limitations, this study proposes an indirect, non-contact monitoring solution using a single AMR sensor to assess current imbalance by measuring the differential magnetic field generated by parallel SiC devices. This method reduces sensor count, minimizes cost, and maintains system integrity.
The core of this approach lies in leveraging the AMR sensor’s ability to detect magnetic field variations in the z-axis direction caused by current flow in parallel PCB traces equivalent to SiC devices. By analyzing the magnetic field distribution around these traces, the current imbalance can be inferred without direct electrical contact. The placement of the AMR sensor is optimized using multiphysics simulation tools like COMSOL to ensure accurate detection. The monitoring system comprises the AMR sensor, signal conditioning circuitry, and a microcontroller unit (MCU) with communication interfaces for real-time data transmission to a backend control center. This enables continuous monitoring of current imbalance in solar inverters, facilitating proactive maintenance and enhanced reliability.
In this article, we detail the design and implementation of the current imbalance monitoring system. We begin by explaining the principles of AMR-based current measurement and the simulation-guided sensor placement. Subsequently, we describe the system architecture, including power supply, signal conditioning, and MCU peripherals. Experimental results from a 1.5 kW solar inverter prototype demonstrate the system’s performance in terms of bandwidth, sensitivity, linearity, and overall effectiveness. The proposed solution offers a cost-effective and reliable means to monitor parallel SiC devices in solar inverters, contributing to the advancement of photovoltaic system monitoring technologies.
AMR Sensor and Optimal Placement for Current Measurement
Anisotropic magnetoresistive (AMR) sensors are advanced magnetic field sensing devices based on semiconductor or alloy materials, characterized by high bandwidth, sensitivity, linearity, and non-contact measurement capabilities. In direct measurement scenarios, the AMR sensor is typically positioned orthogonally to the plane of the measured object. For instance, when measuring current in a PCB trace or the drain-source current of a SiC MOSFET in a TO263-7 package, the sensor is placed to detect the magnetic field perpendicular to the current flow. Since SiC MOSFETs contain an inherent anti-parallel body diode, allowing bidirectional current flow, they can be modeled as PCB traces for analysis purposes. In solar inverters, SiC MOSFETs operate at high switching frequencies, resulting in high-frequency pulsating currents in the equivalent traces. Due to the skin effect at high frequencies, current concentrates near the edges of the trace, altering the magnetic field distribution compared to low-frequency conditions.
The output voltage of the AMR sensor is proportional to the current being measured, with the transimpedance gain \( G_A(s) \) expressed as:
$$G_A(s) = \frac{V_{O(AMR)}}{I_{I(AMR)}} = \frac{k_A}{1 + \frac{s}{F_A}}$$
where \( V_{O(AMR)} \) is the output voltage of the AMR sensor, \( I_{I(AMR)} \) is the current through the sensor, \( k_A \) is the sensitivity coefficient, and \( F_A \) is the cutoff frequency in the sensor’s frequency response. This relationship ensures that the sensor can accurately track high-frequency currents prevalent in solar inverters.
To indirectly monitor current imbalance, the differential magnetic field in the z-axis direction between parallel PCB traces is measured. This approach is cost-effective as the AMR sensor’s range does not need to match the maximum current in the traces. For high-frequency applications, the principle remains valid, with the sensor output reflecting the current imbalance magnitude. The key to precise measurement is identifying the optimal sensor position on the PCB. When both parallel traces carry equal currents, the sensor should output zero voltage at the reference point. Given the symmetrical layout requirements for parallel SiC devices in solar inverters, electromagnetic simulations using COMSOL were employed to determine the best sensor placement.
Simulations modeled two parallel PCB traces carrying currents of equal magnitude at 1 Hz and 10 kHz frequencies. The magnetic field distribution analysis revealed that the magnetic flux density is closest to zero at the midpoint between the traces, separated by distance \( D \). Thus, this midpoint was selected for placing the AMR sensor. The simulation results confirm that this position minimizes the common-mode magnetic field, enhancing the sensitivity to differential currents caused by imbalance. The following table summarizes the magnetic flux density characteristics at different frequencies:
| Frequency | Magnetic Flux Density (Gauss) | Notes |
|---|---|---|
| 1 Hz | ~0 | Ideal for low-frequency balance |
| 10 kHz | ~0 | Sufficient for high-frequency applications |
This positioning strategy ensures that the AMR sensor detects only the imbalance component, improving measurement accuracy. The sensor’s high bandwidth, typically 0–5 MHz, covers the switching frequencies of modern SiC-based solar inverters, making it suitable for real-time monitoring.
System Design for Current Imbalance Monitoring
The current imbalance monitoring system integrates the AMR sensor with signal conditioning circuits and an MCU to process and transmit data. The overall system block diagram includes the power supply, AMR sensor, signal conditioning, MCU with analog-to-digital converter (ADC), and communication interfaces. The input power is derived from the solar inverter’s auxiliary power supply, providing ±5 V. Low-dropout regulators (LDOs) are used to generate stable voltages for different components. LDO1 converts 5 V to 3.3 V for the MCU, while LDO2 produces a clean 3.3 V supply for the AMR sensor and signal conditioning circuits, emphasizing low noise and high power supply rejection ratio (PSRR) to minimize ripple-induced errors.
The signal conditioning circuit amplifies the small output voltage of the AMR sensor, which typically ranges below 100 mV. A pseudo-differential design is adopted to handle the bipolar output of the sensor, centering the signal around a 2.5 V reference. The circuit employs high-speed operational amplifiers (op-amps) with high bandwidth and slew rate to preserve signal integrity at high frequencies. The gain of the signal conditioning stage, \( G_{OA} \), is given by:
$$G_{OA} = (2.5 + V_{AMR+} – V_{AMR-}) \frac{R_X}{R_Y}$$
where \( V_{AMR+} \) and \( V_{AMR-} \) are the positive and negative outputs of the AMR sensor, respectively, and \( R_X \) and \( R_Y \) are resistors that set the amplification ratio. A digital potentiometer for \( R_X \) allows adjustable gain to accommodate varying imbalance levels. The op-amp must feature low input offset voltage, high gain-bandwidth product, and high slew rate to avoid distortion in high-frequency signals.
The MCU is equipped with ADC modules for sampling the conditioned signal and various communication interfaces for data transmission. The ADC should have a high sampling rate and throughput to capture rapid changes in current imbalance. Input protection circuits, such as clamp diodes, are added to prevent overvoltage damage. For connectivity, the MCU supports embedded interfaces like UART, SPI, and I2C, as well as external interfaces like Ethernet and RS-485, enabling seamless integration with the solar inverter’s backend control system. Wireless modules like Wi-Fi or LTE can be used for remote data transmission, facilitating real-time monitoring and control.
The following table outlines the key components and their roles in the monitoring system:
| Component | Function | Key Specifications |
|---|---|---|
| AMR Sensor | Detects magnetic field differential | High bandwidth, sensitivity |
| Signal Conditioning | Amplifies sensor output | High slew rate, low noise |
| MCU | Processes data and communicates | High-speed ADC, multiple interfaces |
| LDO Regulators | Provides stable power | Low noise, high PSRR |
This structured design ensures reliable operation of the monitoring system in the demanding environment of solar inverters, enabling accurate detection of current imbalances in parallel SiC devices.
Component Selection and Experimental Validation
Selecting appropriate components is crucial for the performance of the current imbalance monitoring system. The AMR sensor must exhibit high bandwidth, low offset voltage, adequate sensitivity, and a suitable measurement range. For instance, the HMC1041Z z-axis AMR sensor from Honeywell was chosen due to its 5 MHz bandwidth, sensitivity of 1 mV/V/Gauss, and offset voltage of ±0.5 mV/V. Its range of -6 to 6 Gauss is sufficient for detecting imbalance in solar inverter applications. The sensor’s bandwidth exceeds typical switching frequencies of solar inverters, ensuring accurate measurement without distortion.
For voltage regulation, LDO1 (TPS7A90) provides 3.3 V for the MCU with good noise performance and 500 mA output current. LDO2 (TPS7A94) supplies the AMR sensor and op-amp, featuring ultra-low noise (0.46 μVrms) and high PSRR to maintain signal purity. The op-amp in the signal conditioning stage, such as the ADA4099-1, offers low input offset voltage (30 μV max), high slew rate (4 V/μs), and a gain-bandwidth product of 8 MHz, adequate for amplifying high-frequency signals in solar inverters.
The MCU, STM32F407IE, operates at 168 MHz and includes 12-bit ADCs with a sampling rate of 2.4 MSPS, sufficient for capturing imbalance dynamics. It supports Ethernet, UART, SPI, and I2C interfaces, enabling flexible communication with control systems. The following table summarizes the key parameters of the selected AMR sensor:
| Parameter Name | Typical Value |
|---|---|
| Supply Voltage | 5 V |
| Range | -6 to 6 Gauss |
| Sensitivity | 1 mV/V/Gauss |
| Bridge Offset | ±0.5 mV/V |
| Bandwidth | 5 MHz |
| Hysteresis Error | 0.15% |
| Operating Temperature | -40 to 125 °C |
Experimental tests were conducted on a 1.5 kW solar inverter prototype with an inverter-stage switching frequency of 50 kHz. Parallel SiC MOSFETs (C3M0120065J from Cree) were used, each with an on-state resistance of 120 mΩ and gate charge of 26 nC, suitable for high-frequency power conversion. The monitoring system was integrated into the inverter, and waveforms were captured under different load conditions.
When both SiC MOSFETs carried equal currents, the monitored output voltage \( V_{mon} \) remained near zero, confirming minimal imbalance and correct sensor placement. With one MOSFET shut off and the other carrying the full load current, the system quickly detected the imbalance, as shown by a significant change in \( V_{mon} \). The system’s sensitivity and rapid response were validated, with the output voltage closely tracking the current imbalance \( I_{mis} \). Linearity tests across various load levels demonstrated a strong correlation between \( I_{mis} \) and the op-amp output voltage \( V_{opa} \), though slight non-linearity occurred near zero imbalance due to sensor and op-amp offset voltages.
Data communication to the backend control center was implemented via Ethernet, with hourly measurements of current imbalance absolute values recorded during operational hours. The control center successfully monitored the imbalance trends, enabling real-time awareness of the solar inverter’s condition. For comparison, a traditional monitoring system using two Hall-effect current sensors was also tested. The relative error \( E_r \) between the proposed AMR-based method and the traditional approach was calculated as:
$$E_r = \left| \frac{I_{mis,AMR} – I_{mis,Hall}}{I_{mis,Hall}} \right| \times 100\%$$
Results showed that \( E_r \) remained below 1% across all tests, indicating comparable accuracy. However, the AMR-based system offered advantages in circuit simplicity, reduced component count, and lower cost, making it more suitable for widespread deployment in solar inverters.
Performance Analysis and Discussion
The proposed current imbalance monitoring system for solar inverters exhibits several key performance characteristics. The high bandwidth of the AMR sensor allows it to respond to fast-changing currents in SiC devices, ensuring accurate detection even at high switching frequencies. The sensitivity of the system enables it to detect small imbalance levels, which is critical for preventing localized overheating and enhancing the reliability of solar inverters. The linearity of the output voltage with respect to current imbalance facilitates straightforward calibration and interpretation of results.
In terms of stability, the use of high-PSRR LDO regulators and low-noise op-amps minimizes the impact of power supply variations and environmental noise. The signal conditioning circuit’s pseudo-differential design effectively handles the bipolar nature of the AMR output, maintaining signal integrity. The MCU’s high-speed ADC and communication capabilities ensure timely data acquisition and transmission, supporting real-time monitoring applications.
The following equation models the overall system response, combining the sensor and signal conditioning stages:
$$V_{out} = G_{OA} \cdot G_A(s) \cdot I_{mis} + V_{offset}$$
where \( V_{out} \) is the final output voltage, \( I_{mis} \) is the current imbalance, and \( V_{offset} \) accounts for any residual offset voltages. In practice, \( V_{offset} \) can be calibrated out during system initialization. The table below compares the proposed system with traditional monitoring methods:
| Aspect | Proposed AMR-Based System | Traditional Sensor-Based System |
|---|---|---|
| Sensor Count | Single sensor | Multiple sensors |
| Parasitic Inductance | None (non-contact) | Introduced (contact) |
| Cost | Lower | Higher |
| Accuracy | High (error < 1%) | High |
| Complexity | Simple circuit | More complex |
The experimental results validate the system’s ability to operate effectively in a solar inverter environment. The monitoring system not only detects imbalance but also provides actionable data for maintenance decisions. By integrating with backend control systems, it supports the broader goal of achieving smart, observable, and controllable photovoltaic systems. Future work could focus on enhancing the system’s robustness against electromagnetic interference and extending its application to other power electronic converters beyond solar inverters.
Conclusion
In this study, we have developed and validated a non-contact current imbalance monitoring technique for parallel SiC devices in solar inverters using an AMR sensor. The sensor placement was optimized through COMSOL simulations to detect magnetic field differentials in the z-axis direction. The monitoring system incorporates signal conditioning circuits and an MCU with communication interfaces, enabling real-time data transmission to a backend control center. Experimental tests on a 1.5 kW solar inverter prototype demonstrated the system’s high bandwidth, sensitivity, linearity, and simple circuit structure. The proposed solution offers a cost-effective and reliable alternative to traditional monitoring methods, contributing to improved reliability and maintenance of solar inverters. By facilitating continuous monitoring of current distribution, this approach supports the advancement of photovoltaic system technologies and their integration into modern power grids.