The global transition towards sustainable energy has positioned photovoltaic (PV) power generation as a cornerstone of future electricity supply. To enhance the economic viability and efficiency of PV systems, transformerless solar inverters have gained significant traction. By eliminating the bulky and lossy line-frequency transformer or the complex high-frequency transformer, these systems achieve higher efficiency, reduced size, lower weight, and decreased cost. Among various topologies, the three-level neutral-point-clamped (NPC) inverter is particularly attractive for three-phase grid-connected applications. It offers superior output voltage quality with lower distortion and reduced dv/dt stress on the grid filter and the PV panels themselves. However, the absence of galvanic isolation introduces a critical challenge: the amplification of common-mode (leakage) currents.
This leakage current flows through the inherent parasitic capacitance between the solar PV array and the earth ground. It can lead to increased electromagnetic interference (EMI), elevated ground currents, additional losses, distortion of the grid current, and poses potential safety hazards. Therefore, effective suppression of leakage current is paramount for the reliable and safe operation of transformerless three-phase solar inverters. This article delves into the common-mode characteristics of transformerless three-level NPC solar inverters, presents a modeling approach, analyzes a filtering solution, and addresses the consequential challenge of DC-link neutral-point potential balancing.
Common-Mode Model of a Three-Level Solar Inverter
The structure of a transformerless three-phase three-level PV inverter system is considered, including the PV array’s parasitic capacitance to ground (CPV), the DC-link capacitors (Cdc1, Cdc2), the inverter bridge, and an output LCL filter. The common-mode voltage (Vcm) is defined as the average of the three-phase inverter output voltages with respect to the negative DC bus or the neutral point. The leakage current (icm) is the current flowing from the grid neutral point through the earth connection and the parasitic capacitance back to the PV array.
By applying Kirchhoff’s laws and assuming a symmetrical three-phase grid, the dynamics of the leakage current can be derived. The key insight is that the common-mode circuit is excited by two main sources:
- The high-frequency components of the inverter’s common-mode voltage, Vcm.
- The differential voltage between the two DC-link capacitors, Udm = Udc1 – Udc2.
The common-mode voltage Vcm contains both low-frequency components (e.g., from third-harmonic injection) and high-frequency switching components. The high-frequency components, at the switching frequency and its multiples, are the primary drivers of significant leakage current. The differential DC-link voltage Udm arises from an imbalance in the neutral-point current and also contributes to the common-mode disturbance. A simplified common-mode equivalent circuit can be established, as shown in the analysis, leading to the following governing equation:
$$ V_{cm} = (L + L_g)\frac{di_{cm}}{dt} + \frac{1}{3C_{PV}}\int i_{cm} dt + \frac{1}{2}U_{dm} $$
where L is the inverter-side filter inductance, Lg is the grid-side filter inductance, and CPV is the total PV parasitic capacitance. This model clearly identifies the excitation sources and provides the basis for designing suppression techniques.
Leakage Current Suppression Using an Improved LCL Filter
A highly effective and practical method to attenuate the high-frequency leakage current is the use of a modified LCL filter. In a standard LCL filter for solar inverters, the star point (N) of the three filter capacitors (Cf) is left floating. In the improved configuration, this star point is directly connected to the DC-link midpoint (the neutral point, O). This simple connection creates a low-impedance path for the high-frequency common-mode current components, effectively short-circuiting them before they can excite the PV parasitic capacitance.
The mechanism can be understood by analyzing the modified common-mode circuit. The connection introduces the filter capacitor bank (3Cf in the common-mode path) in parallel with the PV capacitance CPV. The transfer function H(s) from the common-mode voltage Vcm to the voltage across the parasitic capacitor VPV becomes a second-order system. Proper design of the LCL filter parameters (L, Lg, Cf) ensures that the resonant frequency of this common-mode path is placed to attenuate the dominant switching frequency harmonics present in Vcm. The simplified transfer function is given by:
$$ H(s) = \frac{V_{PV}(s)}{V_{cm}(s)} \approx \frac{3}{2(LC_f + L_gC_f + \frac{2}{3}L C_{PV})s^2 + 2} $$
The resonant frequency ωres of this system is always lower than the resonant frequency ωres1 of the original LCL filter’s differential-mode. By designing ωres to be near ωres1, the high-frequency content of VPV is dramatically filtered out. Since the leakage current icm is proportional to dVPV/dt, this results in a substantial reduction of leakage current, often meeting stringent safety and EMI standards.
Impact on Neutral-Point Potential and the Generalized Distribution Factor Algorithm
While the improved LCL filter successfully mitigates leakage current in solar inverters, it fundamentally alters the dynamics of the DC-link neutral-point potential in a three-level NPC inverter. In a standard setup, the neutral-point current inp is solely determined by the inverter switching states and the three-phase output currents. The balance of the two DC-link capacitor voltages is controlled by managing the average neutral-point current to zero over a switching period, typically by adjusting the dwell times of redundant small voltage vectors.
The connection of the filter capacitor star point to the neutral point introduces two critical changes:
- Zero-Sequence Current Injection: A zero-sequence current i0 now flows into the neutral point from the filter capacitors. This current is not present in the standard topology.
- Modified Neutral Current for Redundant Vectors: The current flowing into the neutral point during the application of a switching state is no longer simply the current of the phase(s) clamped to the midpoint. It must account for the injected zero-sequence current i0.
Therefore, the total charge Qnp extracted from the neutral point over a switching period Ts is given by:
$$ Q_{np} = \int_0^{T_s} (i_{np} – i_0) dt $$
where inp is the classical neutral current determined by the switching states. Traditional neutral-point balancing algorithms, such as the popular Distribution Factor (DF) method, do not account for the i0 term and the modified inp values. Applying them directly can lead to poor voltage balancing or even instability.
A Generalized Distribution Factor Algorithm is required. This algorithm still operates by adjusting the share (factor f) between the dwell times of a pair of redundant small vectors (e.g., state 100 and state 211 in Sector I). However, the calculation of the ideal factor f is modified to include the effects of i0 and the corrected inp. The goal remains to make Qnp equal to the charge imbalance corresponding to the existing capacitor voltage difference, Qnp0 = C * Udm.
The corrected neutral currents for the redundant small vector pair (100) and (211) are:
| Vector | Classical inp | Corrected inp (with i0) |
|---|---|---|
| 100 (Positive Small) | iA | iA |
| 211 (Negative Small) | -iA | i0 – iA |
Considering the switching sequence in a sector and solving for the distribution factor f that enforces Qnp = Qnp0 leads to the generalized formula. For a switching sequence involving vectors with dwell times t1, t2 and neutral currents inp1, inp2, the factor f for the redundant pair (with total time t0) is:
$$ f = \frac{2C U_{dm} – i_{np1}t_1 – i_{np2}t_2 + (i_{np0}+i_{np7})t_0/2 – i_0 T_s}{(i_{np0} – i_{np7})t_0} $$
where inp0 and inp7 are the corrected neutral currents for the positive and negative small vectors, respectively. This generalized algorithm precisely controls the neutral-point potential in solar inverters employing the improved LCL filter for leakage current suppression. It seamlessly reduces to the classical DF algorithm when i0 = 0.
Summary of Key Advantages and Implementation
The combination of the improved LCL filter and the generalized distribution factor algorithm presents a comprehensive solution for high-performance, transformerless three-level solar inverters. The key benefits include:
- Effective Leakage Current Attenuation: The filter virtually eliminates high-frequency common-mode voltages across the PV parasitic capacitance.
- Maintained Output Quality: The LCL filter provides excellent attenuation of switching harmonics in the differential-mode grid current.
- Stable Neutral-Point Potential: The generalized control algorithm ensures balanced DC-link capacitor voltages despite the filter modification.
- Practical Implementation: Both the hardware modification (a single wire connection) and the software update (modified control law) are straightforward to implement in digital signal processor (DSP)-based controllers.
For engineers designing three-level solar inverters, the following steps are recommended:
- Model the Common-Mode Path: Estimate the PV array parasitic capacitance CPV (typically 50-150 nF/kW).
- Design the Improved LCL Filter: Select L, Lg, and Cf to meet grid current harmonic standards (e.g., IEEE 1547, IEC 61727) and to place the common-mode resonant frequency for effective attenuation of the switching harmonics.
- Implement the Generalized SVPWM: Integrate the corrected neutral current calculations and the generalized distribution factor formula into the space vector pulse-width modulation (SVPWM) routine of the DSP.
- Validate Performance: Test the system for leakage current (should be within safety limits, e.g., < 300 mA), grid current THD, and DC-link voltage balance under varying power and solar irradiance conditions.
This approach addresses the two most critical challenges in transformerless multilevel solar inverters—safety due to leakage current and reliability due to DC-link imbalance—paving the way for their widespread adoption in next-generation, high-efficiency PV systems.