Current-Sensorless Robust Sliding Mode Control for the Interleaved DC-DC Boost Converter

In the world of modern electronics, power is rarely delivered in a steady, unchanging stream. Devices like electric vehicles, battery chargers, and renewable energy grids rely on DC-DC converters to take a raw voltage from a source and reshape it into a precise, stable level needed by the load. Think of these converters as the heart of a power system, pumping energy through a circuit and regulating the flow to ensure the output remains constant even when the input fluctuates or the demand changes. However, keeping this flow steady is difficult. The components inside these converters are imperfect, the environment introduces noise, and the load can change in an instant. To manage this, engineers traditionally use a combination of sensors to measure the voltage and the current, feeding that data back into a control system that adjusts the flow. But adding sensors increases cost, complexity, and the risk of failure, while the high-frequency switching required to move the power can create electrical noise that confuses the sensors themselves.

A researcher at Northern Illinois University has proposed a way to solve this problem by removing the need for current sensors entirely. Their work focuses on a specific type of power converter called an interleaved boost converter, which uses two parallel paths to handle power more efficiently than a single path could. In a typical setup, engineers would need to measure the current flowing through each of these two paths to ensure they are balanced and to help the controller react quickly to changes. The new method, however, relies on a single measurement: the output voltage. By using a sophisticated mathematical model and a special type of estimator, the researcher created a control system that can infer the behavior of the currents and the health of the entire circuit just by watching the voltage. This approach not only cuts the cost and complexity of the hardware but also makes the system more robust against the inevitable imperfections and variations found in real-world components.

The core of this innovation lies in how the researcher simplified the complex, two-path system into a single, manageable model. Instead of trying to track the individual currents in each of the two parallel legs of the converter, they mathematically transformed the system into an equivalent single-phase representation. This transformation allowed them to group all the messy, unpredictable factors—such as differences between the two legs, changes in the input voltage, and variations in the load—into a single "lumped disturbance." Imagine this disturbance as a hidden force pushing against the system; the researcher designed an observer, a kind of digital detective, that continuously estimates the size and direction of this force without ever needing to see the currents directly. This observer reconstructs the missing state of the system, allowing the controller to compensate for the disturbance and keep the output voltage steady, even when the components are mismatched or the environment is noisy.

To test this idea, the researcher built a detailed simulation of the converter, running it through a series of rigorous scenarios that mimicked real-world challenges. They programmed the system to operate with significant differences between its two parallel legs, using components that were not identical, such as inductors with different values and resistors with varying resistance. In a standard system, such mismatches would cause the currents to split unevenly, potentially damaging the components or destabilizing the output. Yet, in these simulations, the new controller maintained a perfectly stable output voltage. It handled sudden jumps in the load, where the power demand changed instantly, and it withstood fluctuations in the input voltage, all while operating with only the output voltage sensor active. The system did not just survive these conditions; it tracked the target voltage with zero steady-state error, meaning the output settled exactly where it was supposed to be, with no lingering deviation.

The study also examined how the system performed when the measurements themselves were corrupted by noise, a common issue in high-speed electronic circuits. The researcher added a layer of random electrical noise to the voltage signal, simulating the interference that occurs in a real device. Even with this noise, the estimator was able to filter out the high-frequency chatter and focus on the underlying trends of the system. It successfully tracked the average value of the disturbances, ignoring the rapid, chaotic spikes caused by the switching of the power electronics. This filtering capability is crucial because it prevents the control system from overreacting to false signals, which could otherwise cause the voltage to oscillate wildly. The results showed that the controller could maintain tight regulation despite the noise, proving that the method is robust enough for practical applications where perfect signals are impossible to obtain.

One of the most significant findings of this work is that the system does not require the two parallel legs of the converter to be perfectly matched. In many traditional designs, engineers must carefully select components to ensure the two paths are identical, or they must add complex balancing circuits to force the currents to share the load equally. The new method explicitly incorporates the possibility of mismatched components into its mathematical framework. It treats the imbalance not as a failure to be corrected by a separate loop, but as part of the total disturbance to be estimated and compensated. This means the system can function correctly even if one leg has a slightly different inductor or a different resistance than the other. The simulations confirmed that while the currents in the two legs might not be perfectly equal under these mismatched conditions, the total output voltage remained stable, and the overall system performance was not compromised.

The researcher also developed a clear, step-by-step guide for tuning the parameters of their controller and estimator. This is a critical contribution because complex control systems often require difficult, trial-and-error adjustments to work correctly. By linking the tuning parameters to the natural time constants of the system and the switching frequency of the converter, they provided a systematic way to choose the right settings. This ensures that the estimator reacts quickly enough to catch disturbances without becoming so sensitive that it amplifies noise. The simulations demonstrated that with these tuned parameters, the system could recover from disturbances rapidly, settling back to the target voltage in a very short time without overshooting or oscillating. This level of performance, achieved without the extra sensors, suggests a path toward more reliable and cost-effective power electronics for a wide range of applications.

While the results are promising, the author is careful to note that these findings come from computer simulations using both averaged models and detailed switched models of the circuit. The simulations included realistic details such as the resistance of the switches, the forward voltage of the diodes, and the internal resistance of the capacitors, making the test environment very close to reality. However, the work has not yet been tested on physical hardware. The author states that future work will focus on experimental validation to confirm that the system performs as expected in a real-world setting. Until then, the simulations serve as a strong proof of concept, demonstrating that a current-sensorless approach is theoretically sound and practically viable. The study rules out the necessity of measuring input currents or individual phase currents for robust control, showing that a single voltage sensor, combined with a smart estimation algorithm, is sufficient to handle the complexities of an interleaved boost converter.

The implications of this work extend beyond just the specific circuit tested. By showing that complex, multi-phase power systems can be controlled effectively with minimal sensing, the research opens the door to simpler, cheaper, and more reliable power supplies. In fields like electric transportation and renewable energy, where reliability and cost are paramount, reducing the number of sensors can significantly lower the barrier to entry for advanced power management systems. The ability to handle component mismatches and parameter variations without complex balancing loops means that manufacturers could use standard components without needing to sort them for perfect matching, further reducing costs. The robustness against noise and disturbances also suggests that these systems could operate in harsh environments where electrical interference is common.

Ultimately, the paper presents a shift in how engineers might think about controlling power converters. Instead of trying to measure every variable to maintain control, the new approach uses mathematical insight to infer the necessary information from a single, easily measurable signal. It turns the problem of missing data into an opportunity for simplification, using a robust estimation technique to fill in the gaps. The simulations show that this method works, keeping the voltage steady and the system stable even when the components are imperfect and the environment is noisy. As the author moves toward experimental validation, the potential for this technology to transform the design of power electronics becomes increasingly clear, offering a path to more efficient and resilient energy systems for the future.