In my long journey as a power electronics engineer, I have come to deeply appreciate the critical role of reliable inverter architectures, especially when we talk about types of solar inverters. Among the myriad of configurations, the single-phase full-bridge inverter with multiple parallel sub-modules stands out as a robust solution for high-reliability applications, such as those found in marine platforms and, by extension, in large-scale solar photovoltaic systems. In this article, I share my research and insights into the operation strategy and simulation of such a system, focusing on the startup, current-sharing, and shutdown sequences. My goal is to illustrate how different types of solar inverters can benefit from these advanced control strategies, ensuring safe and efficient power conversion.
My work centers on the fact that for high-reliability scenarios, we often use a single-phase inverter composed of multiple output paralleled sub-modules, each operating in hot standby. While much of the literature has addressed current sharing among parallel modules, there remains a gap in the study of the startup timing control. I have therefore proposed a full-process control strategy based on virtual impedance droop control. This strategy not only limits current spikes but also ensures power sharing among modules. By examining various types of solar inverters, I can generalize this approach to many real-world systems.
System Topology and Control Functions
To clearly convey the principles, I take the example of an inverter with two parallel sub-modules, though the concept easily scales to more modules. Each sub-module has its own independent controller. The main circuit topology includes an input circuit breaker, input contactors, pre-charge resistors, power switches (e.g., IGBTs with antiparallel diodes), output filter inductors, and output circuit breakers. The key difference among various types of solar inverters lies in the input source (e.g., PV panels, battery banks) and the output requirements (grid-tied or standalone). However, the internal paralleling structure remains similar.
Each sub-module controller hosts several functional blocks: timing control, background loop, output voltage phase-locked loop (PLL), output current PLL, power calculation, power droop, dual-loop control, and modulation. The timing control module orchestrates the entire sequence from idle to shutdown. The background loop continuously monitors sensor values for fault detection. The PLL modules, based on the second-order generalized integrator (SOGI), extract the phase angle of the voltage and current. The power calculation block computes active and reactive power in the αβ stationary reference frame using the following formulas:
$$
P = \frac{1}{2}(v_\alpha i_\alpha + v_\beta i_\beta)
$$
$$
Q = \frac{1}{2}(v_\beta i_\alpha – v_\alpha i_\beta)
$$
The power droop module employs a P-f and Q-V droop strategy assuming inductive transmission impedance to generate the output voltage reference. The dual-loop controller uses an outer voltage loop and an inner inductor current loop, enhanced with virtual impedance feedback. The modulation block then generates the gate signals for the power switches.
When investigating different types of solar inverters, I found that the control architecture must be flexible to accommodate various power levels and switching frequencies, but the core paralleling technique remains universal.
Timing Control Strategy
The timing control is the heart of my approach. After the input breaker is closed, the controllers enter a standby state. Upon pressing the start button, the input contactor closes, initiating the pre-charge phase. A resister limits the inrush current while the DC-link capacitor charges. Once the voltage across the resistor falls below a threshold, the bypass contactor closes. Then, a random delay (a multiple of 40 ms) is introduced to avoid simultaneous starting of multiple modules. The controller then checks whether the output voltage (after the breaker) is present. If no voltage is detected, the module becomes the first to start and enters the initial output phase. If another module is already running, the new module synchronizes its output voltage PLL and then closes its output breaker at the zero-crossing of the output voltage, minimizing current spikes. Finally, all modules begin PWM output and the droop controller ensures equal power sharing.
I summarize the key steps in the following table for clarity:
| State | Actions | Conditions |
|---|---|---|
| Idle | All contactors open, controller reset | After system power-up |
| Pre-charge | Close input contactor, wait for capacitor charging | Vin > lower limit |
| Bypass | Close bypass contactor when v_R_BP is low | Resistor voltage ≤ threshold |
| Sync & Connect | Random delay, detect output voltage presence; if present, sync PLL and close output breaker at zero-crossing | vo_Aft_rms > threshold |
| Initial Output | Enable all control modules, start PWM at zero-crossing | vo_Aft zero-crossing |
| Normal | Droop control converges power sharing | Steady operation |
| Stop | Disable all modules, stop PWM | Stop button pressed |
This sequence is vital for all types of solar inverters that use parallel configurations, as it prevents destructive inrush currents and ensures seamless hot-swapping.
Simulation Model and Key Waveforms
I built a detailed MATLAB/Simulink model of a two-module inverter to validate the strategy. The system was tested at rated power of 15 kW with a power factor of 0.8 (lagging) and at light load of 1 kW with unity power factor. The simulation results clearly show the effectiveness of the proposed timing control. For the sub-module that starts first (Sub2), the input current spike is limited to 20 A during pre-charge and only 8.2 A when the bypass is engaged. When the second module (Sub1) later connects, it synchronizes with the already active output voltage, so no current spike appears upon breaker closure.
I present the essential output characteristics in the table below, comparing two representative operating points:
| Parameter | Rated Load (15 kW, PF=0.8 lag) | Light Load (1 kW, PF=1) |
|---|---|---|
| Active power sharing difference | 0.06% | 0.8% |
| Reactive power sharing difference | 0.04% | ~50% (small absolute value) |
| Output voltage THD | 2.25% | 0.28% |
| Input current spike (peak) | 20 A | 20 A |
| Output current spike at connection | Negligible | Negligible |
The droop controller ensures that the active and reactive powers converge quickly. At rated load, the difference between the two modules is almost zero. At light load, although the absolute power difference is small (only 4 W), the percentage difference becomes 0.8%, which is still acceptable. The total harmonic distortion (THD) of the output voltage meets the stringent standards of types of solar inverters for marine or grid-tied applications. This demonstrates that the control strategy works well across a wide load range.
In my study, I paid special attention to the transient when a new module joins an already operating system. Without proper timing, high circulating currents can occur. The zero-crossing synchronization, combined with the virtual impedance droop, completely eliminates that risk. This is particularly important for types of solar inverters that need to support hot-plugging for maintenance or redundancy.
Mathematical Foundation of the Droop Control
The power sharing mechanism relies on the well-known droop equations adapted for inverters with inductive output impedance. The frequency droop and voltage droop are given by:
$$
f = f_n – k_p \cdot P
$$
$$
V = V_n – k_q \cdot Q
$$
where \(f_n\) and \(V_n\) are the nominal frequency and voltage, and \(k_p\) and \(k_q\) are the droop coefficients. To improve accuracy, I incorporate a virtual impedance loop in the dual-loop controller. The virtual impedance is implemented by modifying the voltage reference as follows:
$$
v_o^* = v_o – i_o \cdot Z_v
$$
where \(Z_v = R_v + j\omega L_v\) is the virtual impedance. This technique helps to decouple the active and reactive power control and enhances the damping of the system. For different types of solar inverters, the optimal virtual impedance values may differ based on the filter design and cable lengths, but the principle remains the same.
The power calculation mentioned earlier is performed using the αβ transformation. The single-phase quantities are first converted to a two-phase stationary system using a second-order generalized integrator (SOGI). The SOGI transfer function is:
$$
H_{\alpha}(s) = \frac{k \omega_0 s}{s^2 + k \omega_0 s + \omega_0^2}
$$
$$
H_{\beta}(s) = \frac{k \omega_0^2}{s^2 + k \omega_0 s + \omega_0^2}
$$
where \(\omega_0\) is the nominal frequency and \(k\) is the damping factor. This gives us orthogonal components \(i_\alpha\) and \(i_\beta\) from the single-phase current, enabling the use of standard dq control.
The dual-loop controller consists of an outer voltage PI controller and an inner current PI controller. The overall control block can be expressed in the continuous domain as:
$$
G_{v}(s) = k_{pv} + \frac{k_{iv}}{s}
$$
$$
G_{i}(s) = k_{pi} + \frac{k_{ii}}{s}
$$
The output of the voltage controller becomes the current reference for the inner loop, which then generates the modulation index. This cascaded structure provides excellent dynamic response and stability.
Comparison with Other Types of Solar Inverters
When we categorize types of solar inverters, we often think of string inverters, microinverters, and central inverters. However, the parallel sub-module approach I described is most similar to a modular central inverter, where several power units are combined to achieve higher power ratings. In string inverters, multiple strings are connected to a single MPPT, and parallel operation of multiple string inverters on the AC side is common. The timing and droop strategies I have developed are directly applicable to those systems. For microinverters, each unit is dedicated to a single panel, and paralleling is less common, but the principles of droop control still apply for grid synchronization.
Another important types of solar inverters is the hybrid inverter, which can operate in both grid-tied and islanded modes. The proposed strategy is especially valuable for islanded microgrids where multiple inverters must share loads without communication. The virtual impedance droop ensures that even if the inverters have different output impedances (due to cable lengths or manufacturing tolerances), the power is shared proportionally.
I have also tested the strategy under unbalanced loads and nonlinear loads, which are common in practical installations of many types of solar inverters. The results show that the droop controller maintains good voltage regulation and current sharing, though the THD increases slightly under nonlinear loads. Additional filtering or active damping can be employed to mitigate this.
Practical Considerations and Scalability
One of the major challenges in designing parallel inverters is the potential for control conflicts. My timing control ensures that only one module starts first, and subsequent modules synchronize their PLL before injecting current. The random delay prevents multiple modules from attempting to start at exactly the same time, which could cause a race condition. In a system with more than two modules, the logic remains the same: each module independently monitors the output voltage and decides whether to be the first or to synchronize.
For scalability, the droop control parameters must be tuned to match the power rating of each module. If modules have different ratings, the droop coefficients should be inversely proportional to their power capacities. For example, if module A has twice the power rating of module B, then \(k_{pA} = k_{pB}/2\) and \(k_{qA} = k_{qB}/2\). This ensures proportional power sharing among various types of solar inverters with mixed capacities.
Another practical aspect is the selection of the pre-charge resistor. The resistor value must limit the inrush current to a safe level while also allowing the capacitor to charge quickly. In my simulation, a 20 A peak is acceptable, but for larger systems, the resistor may need to be bigger or a two-stage pre-charge may be used. The bypass contactor can be a mechanical relay or a solid-state device, depending on the voltage and current ratings.
The communication-less nature of the droop control makes it highly reliable, as there is no single point of failure. This is a key advantage for critical applications where downtime is unacceptable. In many types of solar inverters, especially those used in off-grid or mission-critical systems, this reliability is paramount.
Conclusion
In this work, I have presented a comprehensive operation strategy for a single-phase inverter with multiple parallel sub-modules. By focusing on the timing control from startup to shutdown, and by employing virtual impedance droop control, I have demonstrated that current spikes can be effectively limited and power sharing can be achieved with high accuracy. The simulation results confirm the effectiveness of the strategy under both rated and light loads. Furthermore, the principles are applicable to a wide range of types of solar inverters, including modular central inverters, string inverters, and hybrid inverters, enhancing their robustness and scalability.
I believe that the detailed mathematical models and simulation insights provided here will assist other engineers in designing and implementing similar paralleling techniques. Future work will involve building a multi-module hardware prototype to validate the concept experimentally. The journey of exploring different types of solar inverters and their control challenges continues, and I hope my contribution helps push the boundaries of reliable power conversion.