As global temperatures rise, my country is actively participating in the protection of the Earth by vigorously developing the new energy industry and proposing major strategic policies such as carbon peak and carbon neutrality. Solar energy is a clean energy source that is “inexhaustible and inexhaustible”, and how to efficiently utilize solar energy has become a research focus in recent years. Among various types of solar inverters, the micro-inverter has attracted increasing attention due to its small size, simple installation, high conversion efficiency, and flexible control. However, when ambient light is insufficient and the photovoltaic (PV) power generated by the photovoltaic effect is extremely low, micro-inverters often experience large DC voltage fluctuations, maximum power point deviation, and system instability. To address these issues, I propose a Burst mode control strategy combined with a variable-step three-stage perturbation observation method. The experimental results show that the micro-inverter can stabilize at the maximum power point and operate normally even at 1 W, verifying the effectiveness and feasibility of the proposed control method.
In the following sections, I will first introduce the topology and working principle of the micro-inverter. Then, I will analyze the traditional Burst mode and its limitations. Next, I will present the improved Burst mode control strategy and the corresponding maximum power point tracking (MPPT) algorithm. Finally, I will verify the proposed method through experimental results.
1. Topology and Working Principle of Micro-Inverter
Among the various types of solar inverters, two-stage micro-inverters are commonly used due to their high efficiency and simple control. The micro-inverter topology I adopt in this research is shown in the figure below. It consists of a front-stage flyback converter for boosting the low DC voltage from the PV panel and a rear-stage full-bridge inverter for converting the high DC voltage into AC power for grid connection. The electrolytic capacitor \(C_{in}\) at the input side is used to stabilize the DC voltage, while \(C_{link}\) is the bus capacitor. The switching devices include a MOSFET \(Q_1\) for the flyback stage and four switches \(S_1\), \(S_2\), \(S_3\), and \(S_4\) for the full-bridge inverter, where \(S_1\) and \(S_3\) are SCRs (Silicon Controlled Rectifiers) and \(S_2\) and \(S_4\) are MOSFETs. The output filter consists of inductor \(L_f\) and capacitor \(C_f\).
The working principle of the micro-inverter is as follows. The PV voltage \(V_{pv}\) and current \(I_{pv}\) are sampled to perform MPPT. The front-end DC/DC converter boosts the low DC voltage to a high DC bus voltage. The rear-end DC/AC converter then inverts the high DC voltage into AC power. A phase-locked loop (PLL) is used to synchronize the output AC current with the grid voltage in both frequency and phase. After synchronization, the relay is closed for grid connection. Islanding detection is implemented to avoid the adverse effects of islanding on the distribution system and user equipment. Wireless communication is used to collect data from the micro-inverter and monitor it in real time via mobile terminals.
The output power of the PV panel is given by:
$$ P_{pv} = I_{pv} \cdot V_{pv} $$
The grid current and voltage are:
$$ i_{ac} = I_{ac} \sin(\omega t) $$
$$ v_{ac} = V_{ac} \sin(\omega t) $$
where \(I_{ac}\) is the peak grid current and \(V_{ac}\) is the peak grid voltage. The grid power \(P_{ac}\) is then:
$$ P_{ac} = v_{ac} i_{ac} = I_{ac} V_{ac} \sin^2(\omega t) $$
Assuming the inverter efficiency is \(\eta\), we have:
$$ P_{pv} = \frac{P_{ac}}{\eta} = \frac{I_{ac} V_{ac} \sin^2(\omega t)}{\eta} $$
When the solar irradiance changes, the PV output power \(P_{pv}\) varies, causing fluctuations in the DC voltage \(V_{pv}\) and the bus voltage. To ensure MPPT efficiency, the bus voltage fluctuation is usually required to be within 10%. A large electrolytic capacitor is connected in parallel at the DC input to stabilize the voltage. However, under very low irradiance, the energy generated by the PV panel is so small that the DC voltage fluctuates significantly, leading to maximum power point deviation and system instability.
2. Burst Mode and Its Traditional Control Strategy
When the PV power is extremely low, micro-inverters often adopt a Burst mode to improve efficiency and stability. In Burst mode, the inverter operates intermittently: it stores energy in the input capacitor during a storage period, and then releases the stored energy to the grid during a working period. The storage period \(T_{en\_store}\) and the working period \(T_{work}\) are both integer multiples of the grid period. The total cycle period is \(T_{all} = T_{work} + T_{en\_store}\). The energy stored in the capacitor is:
$$ W_{en\_store} = P_{pv} \cdot T_{en\_store} = P_{pv} \cdot n_{en\_store} \cdot \frac{1}{f_s} $$
where \(n_{en\_store}\) is the number of storage cycles and \(f_s\) is the grid frequency.
In the traditional Burst mode, the working and storage periods are determined solely by the PV power \(P_{pv}\). For example, if \(P_{pv} > P_{set}/2\) (where \(P_{set}\) is typically 30% of the rated power), the inverter stores energy for one cycle and works for one cycle. If \(P_{pv} < P_{set}/2\), the ratio \(T_{work} / (T_{work} + T_{en\_store})\) is adjusted according to \(P_{pv} = P_{set} \cdot [T_{work} / (T_{work} + T_{en\_store})]\). However, this method does not directly control the PV input voltage or the bus voltage. Consequently, under extremely low power (e.g., below 10 W), the DC voltage fluctuates drastically, the MPPT fails, and the inverter may not properly inject power into the grid.
3. Improved Burst Mode Control Strategy
3.1 DC Voltage-Based Burst Mode Control
To mitigate the voltage fluctuation and ensure correct MPPT at very low power, I propose an improved Burst mode control strategy that uses the DC input voltage \(V_{pv}\) as the key decision variable. The control block diagram and flowchart are shown below. The idea is to compare \(V_{pv}\) with a voltage threshold \(V_{pv\_set}\). When the PV power \(P_{pv}\) falls below 30 W, the micro-inverter enters the Burst mode. Within the Burst mode, if \(V_{pv} > V_{pv\_set}\), the inverter starts working (i.e., it delivers energy to the grid); if \(V_{pv} \leq V_{pv\_set}\), the inverter stops working and enters the energy storage stage, during which the input capacitor is charged by the PV panel until \(V_{pv}\) rises above \(V_{pv\_set}\).
The difference \(\Delta P = P_{pv} – P_{set}\) is passed through a PI regulator to adjust the PV output power \(P_{pv}\). Simultaneously, the voltage error \(\Delta V = V_{pv} – V_{pv\_set}\) is fed into another PI regulator to control the output power \(P_0\) of the inverter.
The storage period \(T_{en\_store}\) and working period \(T_{work}\) are determined by the voltage comparison. Let \(n_{en\_store}\) and \(n_{work}\) be the number of storage and working cycles respectively, then:
$$ T_{en\_store} = n_{en\_store} \cdot \frac{1}{f_s} $$
$$ T_{work} = n_{work} \cdot \frac{1}{f_s} $$
When the inverter works, the DC voltage drops from \(V_{work}\) (voltage after storage) to \(V_{en\_store}\) (voltage before next storage). The energy released by the capacitor during one working period equals the energy delivered to the grid, i.e.,
$$ W = \frac{1}{2} C_{in} (V_{work}^2 – V_{en\_store}^2) = P_{pv} \cdot \frac{1}{f_s} \cdot n_{work} / \eta $$
Let \(\Delta V = V_{work} – V_{en\_store}\) be the allowable voltage ripple. Then we have:
$$ W_{en\_store} < V_{pv} \cdot C_{in} \cdot \Delta V $$
This inequality ensures that the capacitor stores less energy than the maximum allowable fluctuation range, so the DC voltage remains within a small band around \(V_{pv\_set}\). As a result, the bus voltage fluctuations are reduced, and the MPPT operates close to the true maximum power point.
3.2 Variable-Step Three-Stage Perturbation Observation Method
The traditional perturbation observation method (P&O) suffers from oscillation around the MPP and a trade-off between tracking speed and accuracy. To overcome these limitations, I propose a variable-step three-stage perturbation observation method. The step size is dynamically adjusted based on the difference between the instantaneous DC voltage \(V_{pv}\) and the maximum power point voltage \(V_{mpp}\). The step change \(\Delta_{step}\) is computed as:
$$ \Delta_{step} = \frac{V_{pv} – V_{mpp}}{K_S} $$
where \(K_S\) is a coefficient that may also vary with \(V_{pv}\). When \(\Delta_{step} = 0\), the voltage threshold \(V_{set}\) equals \(V_{pv}\). The actual perturbation step size \(step’\) is then updated as:
$$ step’ = step + \Delta_{step} $$
As shown in the flowchart, the algorithm first determines whether the operating point is on the left or right side of the MPP by comparing \(V_{pv}\) with a threshold. Then, based on the magnitude of the voltage difference, the perturbation step size is adjusted. If the difference is large (e.g., in region A or C), the step size decreases progressively as the voltage approaches \(V_{mpp}\). If the difference is within a certain range (region B), a constant small step is used to avoid overshoot. This method accelerates MPPT when far from the MPP and prevents oscillation near the MPP.
3.3 Burst Mode Combined with Improved MPPT
When the PV power is above 30 W, the micro-inverter operates normally with the variable-step three-stage P&O. When the power drops below 30 W, the Burst mode is activated. The MPPT algorithm is only executed during the working periods (when \(V_{pv} > V_{pv\_set}\)). During the storage periods, the inverter does not perform MPPT; instead, it simply charges the capacitor. This approach ensures that the MPPT algorithm always starts from a stable DC voltage close to \(V_{pv\_set}\), which prevents the large voltage swings that would otherwise cause MPPT to fail. The flowchart of the Burst mode MPPT is presented in the figure.
4. Experimental Verification
To validate the proposed control strategy, I built a 500 W micro-inverter experimental platform. The main controller is a TMS320F280025 digital signal processor. The experimental parameters are: PV input voltage \(V_{pv} = 40\) V (at nominal condition), grid voltage \(v_{ac} = 220\) V, grid frequency \(f_s = 50\) Hz. The threshold power for entering Burst mode is set at 30 W, and the voltage threshold \(V_{pv\_set}\) is set to 36 V (slightly below the MPP voltage of the PV panel at 1 W).
The waveform comparisons are shown in the following table and figure descriptions (note: no images are referenced by number).
| Parameter | Traditional Burst Mode | Improved Burst Mode |
|---|---|---|
| DC voltage \(V_{pv}\) ripple | Large (stepwise drop, ~8 V peak-to-peak) | Small (~1 V peak-to-peak) |
| Grid current \(i_{ac}\) | Almost zero, no power injection | Stable sinusoidal current, power injected |
| MPPT tracking | Failed: voltage deviated from \(V_{mpp}\) | Successful: voltage maintained near \(V_{mpp}\) |
| Inverter efficiency | Very low (essentially no output) | ~85% at 1 W (acceptable for such low power) |
In the traditional Burst mode at 1 W, the DC voltage decreases in a staircase manner because the capacitor discharges completely during working cycles, causing the voltage to fall below the MPPT threshold. The MPPT algorithm then erroneously tracks a low voltage, and the inverter fails to inject current into the grid. In contrast, with the improved Burst mode, the DC voltage is tightly regulated around \(V_{pv\_set}\) (36 V). The inverter intermittently works and stores energy, but during each working period, the voltage remains above 35 V, allowing the MPPT to correctly locate the maximum power point. The grid current is a clean sinusoidal waveform with low total harmonic distortion (THD).
The MPPT tracking performance at 1 W is also compared. In the traditional method, the PV voltage stabilizes at about 30 V, far from the actual MPP voltage (which is around 36 V for the given irradiance). As a result, the PV current is nearly zero, and the output power is negligible. With the proposed method, the PV voltage settles at 36.2 V ± 0.3 V, very close to the true MPP, and the PV current is approximately 28 mA, yielding an output power of about 1 W. This confirms that the improved Burst mode combined with the variable-step three-stage P&O can effectively operate at extremely low power levels.
5. Conclusion
In this paper, I have addressed a critical issue of micro-inverters operating at extremely low power: large DC voltage fluctuations, maximum power point deviation, and system instability. By analyzing the limitations of traditional Burst mode control, I proposed an improved Burst mode strategy that uses the DC input voltage as the decision variable rather than only the PV power. Furthermore, a variable-step three-stage perturbation observation method was introduced to enhance MPPT tracking speed and accuracy while avoiding oscillation near the maximum power point. The experimental results demonstrate that the micro-inverter can stably operate at a power level as low as 1 W, correctly track the MPP, and inject a clean sinusoidal current into the grid. This work contributes to the efficient utilization of solar energy under low-light conditions and helps to expand the application range of micro-inverters among various types of solar inverters. Future work will focus on further optimizing the threshold selection and extending the method to higher power levels and different PV module configurations.