The evolution of photovoltaic (PV) systems has led to diverse architectures, among which the AC module concept, utilizing a dedicated solar inverter per panel, offers significant advantages in scenarios with partial shading and flexible expansion. These micro-inverters, typically rated between 200W and 300W, face the distinct challenge of converting a wide, low DC input voltage (e.g., 20-50V) to a stable high-voltage DC bus for grid inversion, all while maintaining high conversion efficiency across the entire operating range. This article presents a comprehensive study on a two-stage micro-inverter topology, focusing on an innovatively controlled front-end DC-DC stage that achieves wide-input-range, high-efficiency power conversion.
The proposed system architecture for the micro-power solar inverter is depicted in the conceptual block diagram. It consists of two primary power conversion stages. The first stage is an isolated DC-DC converter based on a full-bridge LLC resonant topology with a voltage-doubler rectifier. This stage is responsible for boosting the highly variable PV panel voltage to a stable high-voltage DC bus (e.g., 380 VDC). The second stage is a standard full-bridge inverter that converts this DC bus voltage into a sinusoidal AC current synchronized with the utility grid. The control system is orchestrated by a digital signal processor. The front-end LLC controller regulates the bus voltage using a novel hybrid control law. The MPPT algorithm, implemented by perturbing the output power reference of the inverter stage, generates the current amplitude command. This command, along with phase information from a PLL, is used by a deadbeat current controller to produce high-quality grid current.
The core innovation lies in the control strategy for the LLC resonant solar inverter stage. Traditional frequency modulation (PFM) control, while efficient, necessitates an impractically wide switching frequency range for wide input voltage applications. Alternative phase-shift (PWM) control can limit frequency variation but incurs high circulating current losses at light loads and high input voltage. The proposed hybrid control strategy optimally combines both techniques. The operation is divided into two regions based on the switching frequency \(f_s\) relative to the resonant frequency \(f_r\):
- PFM-Only Region (\(f_s \leq f_r\)): When a high voltage gain is required (low input voltage), pure frequency modulation is employed.
- Hybrid PFM-PWM Region (\(f_s > f_r\)): When a lower gain is needed (high input voltage), the controller employs both PFM and PWM simultaneously. This approach effectively caps the maximum required switching frequency and minimizes circulating currents compared to pure phase-shift control at high input voltages.
The voltage gain \(M\) of the full-bridge LLC with voltage-doubler under hybrid control can be derived using the Fundamental Harmonic Approximation (FHA):
$$
M(f_s, D, k, Q) = \frac{\sin\left(\frac{D\pi}{2}\right)}{\sqrt{ \left(1 – \frac{1}{f^2}\right)\left(\frac{1}{k} + 1\right)^2 + \left(f – \frac{1}{f}\right)^2 \left[1 – \cos(D\pi)\right] Q^2 }}
$$
where:
- \(f = f_s / f_r\) is the normalized frequency,
- \(D\) is the phase-shift duty ratio (\(0 < D \leq 1\)),
- \(k = L_m / L_r\) is the inductance ratio,
- \(Q = \sqrt{L_r / C_r} / R_{ac}\) is the quality factor, and \(R_{ac}\) is the equivalent AC load resistance.
When \(D = 1\), the equation simplifies to the classic PFM-only LLC gain formula. The relationship between \(f\), \(D\), and \(M\) for a fixed \(k\) is complex, but analysis shows that the hybrid strategy allows for multiple {\(f, D\)} pairs to achieve the same gain \(M < 1\), providing the flexibility to optimize for frequency range or loss minimization. The control algorithm flowchart involves sampling the output voltage, calculating an error, and performing a PI compensation to update the switching frequency. If the new \(f_s\) exceeds \(f_r\), a phase-shift angle is introduced based on a predefined base value and an adjustment proportional to the frequency error.
For the maximum power point tracking in this solar inverter, a perturb-and-observe (P&O) algorithm is applied at the inverter stage. Instead of perturbing the PV voltage or the LLC converter’s operating point (which has multiple control variables \(f\) and \(D\)), the reference amplitude for the grid current is perturbed. The corresponding change in output power is observed to determine the next perturbation direction. This method simplifies the MPPT implementation by decoupling it from the complex dynamics of the LLC stage, relying on the bus voltage regulator to maintain a stable DC link.
The design of the resonant tank parameters (\(L_r\), \(C_r\), \(L_m\)) is critical for optimizing the performance of the LLC-based solar inverter. The goal is to achieve the required voltage gain range while minimizing total losses. Loss analysis indicates that both the RMS current \(I_{Q\_rms}\) and the turn-off current \(i_{Q\_off}\) of the primary switches decrease with increasing \(k\) and \(Q\), suggesting lower conduction and switching losses. However, the achievable maximum gain \(M_{max}\) under PFM control decreases as \(k\) and \(Q\) increase, which would force a wider frequency range. Therefore, a systematic design procedure is required, constrained by two key conditions:
1. Gain Requirement: The converter must provide the maximum required gain \(M_{max}\) at the minimum allowable switching frequency \(f_{min}\):
$$ M(f_{min}, D=1, k, Q) \geq M_{max} $$
2. ZVS Condition: To ensure Zero-Voltage Switching (ZVS) of the primary MOSFETs, the input impedance of the resonant network must be inductive at the operating point. This condition is met when the imaginary part of the input impedance is positive, which translates to:
$$ f_{min} > f_{Im\_0} $$
where \(f_{Im\_0}\) is the normalized frequency where \(Im(Z_{in}) = 0\), given by:
$$ f_{Im\_0} = \sqrt{ \frac{ k^2 Q^2 – k – 1 + \sqrt{(k^2 Q^2 – k – 1)^2 + 4k^2 Q^2 } }{ 2k^2 Q^2 } } $$
The optimal design point is found by identifying the {\(k, Q\)} pair within the feasible region satisfying both constraints that maximizes the product \(k \cdot Q\). Maximizing this product leads to a larger magnetizing inductance \(L_m\), which reduces the circulating current and associated core/cu losses in the transformer.
| Parameter | Value |
|---|---|
| Rated AC Output Power | 250 W |
| Grid Voltage / Frequency | 220 VAC / 50 Hz |
| PV Input Voltage Range (MPPT) | 22 – 45 VDC |
| Nominal Input Voltage | 31 VDC |
| DC Bus Voltage | 380 VDC |
| LLC Resonant Frequency (\(f_r\)) | 65 kHz |
| Switching Frequency Range (\(f_s\)) | 40 – 130 kHz |
| Inverter Stage Switching Frequency | 40 kHz |
| Grid Filter Inductance (L1, L2) | 2.5 mH |
| Parameter | Symbol | Value |
|---|---|---|
| Inductance Ratio | \(k = L_m/L_r\) | 3.7 |
| Quality Factor (at full load) | \(Q\) | 0.4 |
| Resonant Inductance | \(L_r\) | 5.68 μH |
| Resonant Capacitance | \(C_r\) | 1.05 μF |
| Magnetizing Inductance | \(L_m\) | 21.02 μH |
| Transformer Turns Ratio (Np:Ns) | – | 7 : 44 |
Based on the specifications in Table 1 and the design methodology, the resonant tank parameters were calculated and are listed in Table 2. A 250W prototype was built and tested. Experimental waveforms confirmed the operational principles: at nominal input voltage (31V), the LLC stage operated at the resonant frequency (≈65 kHz). At a higher input voltage (39V), the hybrid control was active, with \(f_s > f_r\) and a clear phase-shift observable in the bridge voltages. ZVS was achieved for the primary MOSFETs across the input range, and ZCS was maintained for the secondary-side diodes.
The efficiency of the LLC DC-DC stage was measured across the operational range. The converter maintained an efficiency above 96.0% throughout the MPPT voltage range (25-39V), with a peak efficiency of 97.7%. This demonstrates the effectiveness of the hybrid control strategy in minimizing losses. The complete solar inverter system, including the inverter stage and auxiliary power, achieved a peak efficiency of 95.5%. The grid current quality was excellent, with Total Harmonic Distortion (THD) below 2.0% at both half and full load. The MPPT algorithm demonstrated fast convergence and a tracking accuracy exceeding 99% under standard test conditions.
In conclusion, this work presents a viable and efficient solution for micro-power photovoltaic applications. The proposed two-stage solar inverter architecture, centered on an LLC resonant converter with a novel hybrid PFM-PWM control strategy, successfully addresses the key challenges of wide input voltage range and high conversion efficiency. The hybrid control intelligently limits frequency variation and reduces circulating current losses. Coupled with a simple yet effective output-power-based MPPT algorithm and a systematic resonant tank design methodology, the developed prototype validates the approach, meeting performance targets for efficiency, power quality, and MPPT accuracy. This design contributes to advancing the technology for efficient and reliable AC module systems.