The pursuit of higher efficiency, reduced size, and lower cost in distributed photovoltaic (PV) power generation systems has driven the widespread adoption of transformerless utility interactive inverters. By eliminating the bulky and lossy line-frequency transformer, these systems achieve significant advantages. However, the absence of galvanic isolation introduces a critical challenge: the generation of common-mode leakage currents. These currents flow through the parasitic capacitance between the PV array and ground due to the high-frequency common-mode voltage generated by the inverter’s switching actions. Leakage currents increase grid current harmonics, degrade power quality, and raise serious safety concerns, necessitating reliable suppression techniques.
Various topological families have been developed to address this issue. Half-bridge topologies connect the grid neutral point directly to the midpoint of the DC bus, clamping the voltage across the parasitic capacitor. While effective, they exhibit a pronounced step-down characteristic, requiring approximately twice the input voltage of a full-bridge counterpart for the same output, making them less suitable for many low-voltage applications. Full-bridge topologies using unipolar sinusoidal Pulse Width Modulation (PWM) can also suppress leakage current but inherently generate high-frequency common-mode voltage, leading to larger grid current harmonics and typically requiring two filter inductors, increasing cost. A more fundamental approach is the common-ground topology, which physically connects the grid neutral point to the negative pole of the DC bus. This connection effectively short-circuits the PV parasitic capacitance to ground, theoretically eliminating the path for leakage current altogether.
In this article, I propose a novel common-ground transformerless utility interactive inverter topology based on a switched-capacitor structure. My design aims to completely eliminate leakage current while maintaining a simple structure with a reduced semiconductor count and a straightforward modulation strategy, enhancing its practicality for real-world applications. The core of this utility interactive inverter is its inherent common-ground connection.
Proposed Topology and Operational Principle
The proposed common-ground transformerless utility interactive inverter circuit is illustrated below. It consists of five power switches (S1 to S5), a DC-link stabilizing capacitor (Cdc), a flying or intermediate storage capacitor (Cr), and a filter inductor (Lf) connecting the output to the grid voltage ug. The defining feature is the direct connection of the grid neutral point (N) to the negative rail of the DC input. The PV panel’s parasitic capacitance to ground is represented by CPV.
The operation of this utility interactive inverter is governed by a unipolar PWM strategy designed to minimize switching loss and simplify control. Based on the direction of grid voltage and current, one grid cycle is divided into four operational intervals. Each interval comprises two stages: an energy transfer stage and a freewheeling stage, leading to a total of eight distinct switching modes (M1 to M8). The modulation is designed following key principles: 1) Prefer direct energy transfer between the PV source and the grid to minimize losses; 2) Minimize the number of simultaneously conducting switches to reduce conduction loss; 3) Minimize the switching frequency of devices to reduce switching loss.
The switching states for all five switches across the eight modes are summarized in Table 1. ‘1’ indicates the switch is ON, and ‘0’ indicates it is OFF.
| Interval (ig, ug) | Mode | S1 | S2 | S3 | S4 | S5 |
|---|---|---|---|---|---|---|
| I (+, +) | M1 (Transfer) | 1 | 0 | 1 | 0 | 1 |
| I (+, +) | M2 (Freewheel) | 1 | 0 | 0 | 1 | 1 |
| II (+, -) | M3 (Transfer) | 0 | 1 | 0 | 1 | 0 |
| II (+, -) | M4 (Freewheel) | 1 | 0 | 0 | 1 | 1 |
| III (-, -) | M5 (Transfer) | 0 | 1 | 0 | 1 | 0 |
| III (-, -) | M6 (Freewheel) | 1 | 0 | 0 | 1 | 1 |
| IV (-, +) | M7 (Transfer) | 1 | 0 | 1 | 0 | 1 |
| IV (-, +) | M8 (Freewheel) | 1 | 0 | 0 | 1 | 1 |
The working principles can be categorized into four key patterns (A to D):
- Pattern A (Grid absorbing power, positive half-cycle): During the transfer stage (M1), energy flows directly from the PV source through S3 to the grid and Lf. In the freewheeling stage (M2), Lf freewheels through S4 and the body diode of S5, while the PV source charges the flying capacitor Cr through S1.
- Pattern B (Grid supplying power, negative-to-positive transition): During transfer (M3), energy from the grid charges Cr via S2. During freewheeling (M4), the grid current freewheels through Lf, S4, and the body diode of S5.
- Pattern C (Grid absorbing power, negative half-cycle): During transfer (M5), energy from Cr is delivered to the grid via S2. During freewheeling (M6), Lf freewheels through S4 and S5’s diode, and the PV source charges Cr via S1.
- Pattern D (Grid supplying power, positive-to-negative transition): During transfer (M7), energy flows from the grid directly back to the PV source via S3. During freewheeling (M8), Lf freewheels, and the PV source charges Cr.
This cyclic operation of charging and discharging Cr enables proper power flow and output waveform generation in this utility interactive inverter.
Common-Mode Behavior Analysis
The superior leakage current suppression capability of this utility interactive inverter stems from its excellent common-mode behavior. Common-mode (VCM) and differential-mode (VDM) voltages are defined with respect to the ground point N:
$$
V_{CM} = \frac{V_{AN} + V_{BN}}{2}, \quad V_{DM} = V_{AN} – V_{BN}
$$
where VAN and VBN are the potentials at the AC output terminals A and B relative to N. Consequently:
$$
V_{AN} = V_{CM} + \frac{V_{DM}}{2}, \quad V_{BN} = V_{CM} – \frac{V_{DM}}{2}
$$
To analyze high-frequency common-mode behavior, the grid voltage source can be considered a short circuit due to the much lower grid frequency compared to the switching frequency. Applying the “short-circuit principle” for common-mode analysis, the differential-mode loop and components can be neglected. The high-frequency common-mode equivalent circuits for the energy transfer and freewheeling stages are derived.
For the energy transfer stage (e.g., Mode M1), the equivalent circuit shows that VAN is connected to the positive DC rail (Udc) and VBN is connected to the negative rail (0V). Therefore:
$$
V_{AN} = U_{dc}, \quad V_{BN} = 0
$$
The high-frequency common-mode voltage VCM,HF is:
$$
V_{CM,HF} = \frac{V_{AN} + V_{BN}}{2} = \frac{U_{dc} + 0}{2} = \frac{U_{dc}}{2}
$$
Importantly, this is a constant DC value, not a high-frequency switching voltage.
For the freewheeling stage (e.g., Mode M2), both output terminals A and B are connected to the negative DC rail through conducting paths. Thus:
$$
V_{AN} = 0, \quad V_{BN} = 0
$$
This yields:
$$
V_{CM,HF} = \frac{0 + 0}{2} = 0
$$
Throughout the entire switching cycle, the high-frequency component of VCM is either zero or a constant DC value. The low-frequency common-mode voltage caused by the grid voltage itself is also a constant due to the common-ground connection. Since the parasitic capacitor CPV is connected between the PV positive terminal and ground, and its negative terminal is fixed at ground potential, the voltage across it (VCPV) lacks a high-frequency alternating component. The leakage current ileak is governed by:
$$
i_{leak} = C_{PV} \frac{dV_{CPV}}{dt}
$$
With dVCPV/dt ≈ 0 at switching frequency, the leakage current is effectively eliminated. This analysis confirms the intrinsic low common-mode behavior of the proposed common-ground utility interactive inverter.
Power Loss and Efficiency Modeling
A comprehensive efficiency analysis is crucial for evaluating the performance of any utility interactive inverter. The total power loss Ploss in the proposed topology comprises switch losses, diode losses, and capacitor Equivalent Series Resistance (ESR) losses.
$$
P_{loss} = P_{switch} + P_{diode} + P_{C,ESR}
$$
1. Switch Losses (Pswitch): For the IGBTs and MOSFETs, losses include conduction loss (Pcond) and switching loss (Psw).
Conduction loss for a switch is calculated by integrating its instantaneous loss over the grid period T:
$$
P_{cond} = \frac{1}{T} \int_0^T [V_{ce,sat} + i(t) \cdot R_{ds(on)}] \cdot i(t) \cdot D(t) \, dt
$$
where Vce,sat is the saturation voltage, i(t) is the conducted current, Rds(on) is the on-state resistance, and D(t) is the duty cycle.
Switching loss is approximated using energy loss per switching event:
$$
P_{sw,on} = E_{on} \cdot f_{sw} \cdot \frac{U_{dc}}{U_{dc0}}, \quad P_{sw,off} = E_{off} \cdot f_{sw} \cdot \frac{U_{dc}}{U_{dc0}}
$$
where Eon/Eoff are turn-on/off energy values from datasheets, fsw is switching frequency, Udc is actual DC voltage, and Udc0 is the datasheet test voltage.
Total switch loss: Pswitch = Pcond + Psw,on + Psw,off.
2. Diode Losses (Pdiode): Losses in the body diodes or anti-parallel diodes include conduction loss (PF) and reverse recovery loss (PRR).
$$
P_{F} = V_F \cdot I_{F,avg} + R_F \cdot I_{F,rms}^2
$$
$$
P_{RR} = E_{rr} \cdot f_{sw} \cdot \frac{U_R}{U_{R0}}
$$
where VF is forward voltage, IF,avg/rms are average/RMS currents, RF is dynamic resistance, Err is reverse recovery energy, UR is blocking voltage, and UR0 is the datasheet test voltage. Total diode loss: Pdiode = PF + PRR.
3. Capacitor ESR Losses (PC,ESR): Losses in capacitors Cdc and Cr due to their ESR (rC).
$$
P_{C,ESR} = r_{C} \cdot I_{C,rms}^2
$$
where IC,rms is the RMS current through the capacitor.
The overall efficiency η of the utility interactive inverter is then:
$$
\eta = \frac{P_{out}}{P_{out} + P_{loss}} \times 100\%
$$
where Pout is the output power delivered to the grid.
Based on component parameters (e.g., 400V input, 220Vrms output, 20kHz switching frequency), the calculated efficiency curve for the proposed topology shows a characteristic shape. Efficiency initially increases with output power as conduction losses become more dominant relative to fixed losses, peaks near the designed operating point (e.g., around 400W for a 1kW-rated design), and then gradually decreases at higher power due to increased ESR and conduction losses. This confirms the suitability of this utility interactive inverter for typical residential PV power levels.
Simulation and Experimental Verification
To validate the theoretical analysis of the proposed common-ground transformerless utility interactive inverter, both simulation and a 1kW hardware prototype were developed based on the parameters listed in Table 2.
| Parameter | Value |
|---|---|
| Input DC Voltage (Udc) | 400 V |
| Grid Voltage (ug) | 220 Vrms, 50 Hz |
| Rated Power (Pout) | 1 kW |
| DC-link Capacitor (Cdc) | 330 µF |
| Flying Capacitor (Cr) | 100 µF |
| Parasitic Capacitance (CPV) | 100 nF |
| Filter Inductor (Lf) | 2 mH |
| Switching Frequency (fsw) | 20 kHz |
Simulation Results: Simulation waveforms confirm the key features. The grid current is sinusoidal and in phase with the grid voltage. Most importantly, the common-mode voltage VCM measurement shows essentially no high-frequency content, remaining at a constant DC level throughout the cycle. Consequently, the leakage current is negligible, with a simulated peak value in the low milliampere range, effectively eliminated.
Experimental Results: The experimental waveforms from the 1kW prototype align closely with simulations. The grid voltage and current waveforms demonstrate stable operation with low distortion. A Fast Fourier Transform (FFT) analysis of the grid current reveals a Total Harmonic Distortion (THD) of approximately 2.1%, which complies with stringent grid interconnection standards such as IEEE 1547 and IEC 61727. This low THD is a direct benefit of the eliminated leakage current interference.
The experimental measurement of the leakage current itself shows a peak value of about 12 mA and an average near 10.98 mA. This extremely low value, essentially at the noise level of measurement equipment, conclusively verifies the topology’s ability to eliminate leakage current in a transformerless utility interactive inverter system. The common-ground connection performs its function as designed.
Conclusion
In this work, I have presented the analysis and verification of a novel switched-capacitor based common-ground transformerless utility interactive inverter. The topology’s salient feature is the direct connection of the grid neutral to the DC bus negative rail, which clamps the voltage across the PV parasitic capacitance. A detailed operational analysis using a unipolar PWM strategy demonstrates its working modes. The common-mode analysis proves that this structure inherently produces no high-frequency common-mode voltage, leading to the complete elimination of ground leakage current—a major advantage for safety and power quality. Loss modeling indicates high efficiency, particularly at the optimal power range. Both simulation and experimental results on a 1kW prototype validate the theoretical claims, showing sinusoidal grid current with low THD and negligible leakage current. This utility interactive inverter presents a compelling, simple, and effective solution for safe and efficient transformerless PV grid connection. Future work may focus on extending this common-ground principle to topologies with buck-boost capability to widen the input voltage range for broader application of such utility interactive inverters.