**4. Experimental Results**

The proposed grid-connected micro-inverter (Figure 11) was designed to operate at the rated power 320 W, *VIN* = 25~52 VDC, *IIN.max* = 12 ADC, and *fs* = 60~90 kHz. The grid voltage was 220 Vrms, the grid frequency was 60 Hz, and grid current supplied by the proposed micro-inverter is 0~1.45 Arms. The proposed micro-inverter was implemented using the circuit parameters given in Table 2. The microcontroller used was a MN103DF35 (PANASONIC). For the PI controller in the main controller, *KP* and *KI* were experimentally optimized and set to 9.5 and 200, respectively. The sampling frequency for analog signals is 20 kHz, and the resolution of the analog-to-digital converter is 12 bits. The turns ratio of *LB* is 10:10 and that of *T*1 is 6:19. The resonant frequency *fr* = 35.5 kHz from *Llk* = 100 μH and *C*1 = *C*2 = 100 nF. The MOSFET package of *S*1–*S*4 is PG-TDSON-8 and that of *S*5–*S*8 is D2PAK. Capacitors *Cs*, *C*1 and *C*2 are MPP-film type. The fabricated micro-inverter was compact and slim with 60-mm width, 310-mm length, and 30-mm height.

the

proposed

micro-inverter.



 of

Photograph

**Figure 11.**

Instead of an actual PV module, the photovoltaic simulator ETS600X14CPVF TerraSAS from AMETEK was used as an input source. The solar cell *I*-*V* characteristic curve for the experiment was based on that of the NeON®2 PV module from LG electronics.

Gate-source and drain-source voltages were obtained for *S*1 and *S*2 at *D* ≤ 0.5 (Figure 12a) and at *D* > 0.5 (Figure 12b) at *VIN* = 34 V and *vgrid* = 220 Vrms /60 Hz. The drain-source voltage v*DS*1 of *S*1 drops to 0 V before the gate signal *vS*1 is applied, so *S*1 turns on with ZVS. *S*2 is complementary to *S*1

and achieves a ZVS turn-on. The operation of *S*3 and *S*4 is out of phase with that of *S*1 and *S*2, so *S*3 and *S*4 can also achieve the ZVS turn-on.

**Figure 12.** Gate-source and drain-source voltages of *S*1 and *S*3 for (**a**) *D* ≤ 0.5 and (**b**) *D* > 0.5.

Waveforms (Figure 13) were obtained for *vgrid* and *ig* at *VIN* = 34 V and *vgrid* = 220Vrms / 60 Hz for output power *Po* = 320 W and 64 W. To maximize efficiency, the proposed micro-inverter operates in normal mode at *Po* ≥ 110 W and in AB control mode at *Po* < 110 W. The boundary of the output power at which the proposed micro-inverter switches from the normal mode to AB control mode and vice versa is selected to be in a range where the peak value of *ig* does not exceed the rated grid current.

**Figure 13.** Grid voltage and current waveforms.

Waveforms were obtained for the fixed-frequency (Figure 14a) and the variable-switchingfrequency (Figure 14b) controls. Gate signals of *S*1 and *S*3, *ig* and *vgrid* were measured at *VIN* = 34 V, *vgrid* = 220 Vrms / 60 Hz, and output power *Po* = 320 W. When fixed-switching frequency control was used, *ig* was distorted near zero-crossing, and THD was increased to 5.79%. In contrast, when variable switching frequency control was used, the distortion of *ig* was improved near zero-crossing, and THD was reduced to 2.65%, which is below the requirement for distributed power. The switching frequency *fs* decreased as *ig* increased, so switching loss was also reduced.

Δ*VIN* is higher when conventional burst control is used (Figure 15a) than when AB control is used (Figure 15b), because AB control reduces the energy supplied by *CIN* during one ON-state period. At *VIN* = 34 V, *vgrid* = 220 Vrms / 60 Hz, and *Po* = 32 W, Δ*VIN* was 4.2 V when conventional burst control was used, but 2.4 V when AB control was used.

*Energies* **2019**, *12*, 1234

**Figure 14.** Gate signals of *S*1 and *S*3, grid voltage and grid current in (**a**) the fixed and (**b**) the variable switching frequency controls.

**Figure 15.** Input ripple voltage in (**a**) the conventional and (**b**) the advanced burst controls.

The MPPT efficiency of the proposed micro-inverter was measured (Figure 16) in the range of irradiance from 50 W/m<sup>2</sup> (*Po* = 16 W)–1000 W/m<sup>2</sup> (*Po* = 320 W). In the proposed control scheme, for *Po* < 110 W (burst mode), the MPPT efficiency was kept >95% because Δ*VIN* and Δ*Ig\_ref* are reduced. However, in the conventional control scheme, the MPPT efficiency was reduced to ~88% because fluctuation of *Ig*\_*ref* increased. During burst mode, the maximum MPPT efficiency was >99% for the proposed control scheme but <97.5% for the conventional control scheme.

**Figure 16.** MPPT efficiency depending on control methods.

In a micro-inverter, one of the most important factors is the power conversion efficiency ï*e* for 50~75% load under actual solar irradiation. Therefore, the California Energy Commission (CEC) weighted efficiency to represent this fact has been widely used to measure the performance of micro-inverters. The power conversion efficiency ï*e* (Figure 17) was measured for the proposed micro-inverter; the result indicate that the CEC weighted efficiency [17,18] is 95.55%, in which <sup>ï</sup>*e*(10%) = 91.71%, <sup>ï</sup>*e*(20%) = 94.42%, <sup>ï</sup>*e*(30%) = 95.28%, <sup>ï</sup>*e*(50%) = 96.06%, <sup>ï</sup>*e*(75%) = 95.8%, and <sup>ï</sup>*e*(100%) = 95.72%. The maximum ï*e* is 96.06% for *Po* = 160 W.

**Figure 17.** Power conversion efficiency ï*e* measured at *VIN* = 34 V and *vgrid* = 220 Vrms/60 Hz.
