Design methodology for a photovoltaic emulator based on power hardware-in-the-loop with Opal-RT simulator Authors: Hoai Phong Nguyen, Thuan Thanh Nguyen Published: 2026 ABSTRACT This study introduces a design methodology for a photovoltaic (PV) power emulator system. The proposed framework establishes a simple and rapid approach to constructing a PV emulator by utilizing a real-time simulator to control a power amplifier supply, forcing it to mimic the operational characteristics of a real solar array. This platform effectively serves for the experimental validation of PV characteristic measurements using a resistive load in student laboratory experiments, as well as the performance testing of solar power converters, such as solar MPPT chargers. Furthermore, the proposed method constructs an accurate PV system model operating under static environmental profiles (uniform irradiance and ambient temperature) as well as mismatched series cells under non-uniform irradiance, such as partial shading scenarios. The distinctive features of this methodology are as follows:• Photovoltaic (PV) modeling.• Development of a real-time PV array simulation model.• Real-time power emulation of PV profiles based on a Power Hardware-in-the-Loop (PHIL) platform, utilizing an OPAL-RT OP5707 real-time simulator and a 1-kW APS1000 power amplifier. Specifications table Subject areaEngineeringMore specific subject areaPV emulatorName of your methodA Photovoltaic System Emulation Method Based on PHIL and OPAL-RTName and reference of original methodNot ApplicableResource availabilityHardware: OPAL-RT Real-Time Digital Simulator OP5707 and APS1000 Amplifier Power Supply.Software: MATLAB / Simulink RT-LAB software suite.Source code repository:https://github.com/phongnh3289/PV-modeling Subject areaEngineeringMore specific subject areaPV emulatorName of your methodA Photovoltaic System Emulation Method Based on PHIL and OPAL-RTName and reference of original methodNot ApplicableResource availabilityHardware: OPAL-RT Real-Time Digital Simulator OP5707 and APS1000 Amplifier Power Supply.Software: MATLAB / Simulink RT-LAB software suite.Source code repository:https://github.com/phongnh3289/PV-modeling Background Solar energy is a pivotal renewable energy source that plays a crucial role in sustainable development, reducing carbon dioxide emissions, and advancing global efforts toward net-zero targets [1,2]. Photovoltaic (PV) systems utilize solar modules comprised of numerous series-connected PV cells, where each cell acts as a current source whose output depends directly on solar irradiance and the underlying semiconductor material. When environmental conditions exhibit temporal variations or spatial non-uniformity (such as changes in irradiance and temperature), the resulting fluctuations in current and voltage introduce significant challenges to the control and operation of PV power systems. To validate research on PV control strategies or evaluate commercial devices like solar inverters and solar chargers, PV emulators are commonly deployed. These emulators replicate diverse environmental profiles and power outputs for testing purposes without relying on physical solar arrays [[3], [4], [5]]. Several research approaches have been proposed for PV emulator design. Some studies have focused on achieving fast dynamic response and high stability through simulation by applying intelligent control methods such as Neural Networks and Fuzzy Logic [[6], [7], [8], [9]]. However, these approaches are challenging to implement in practical hardware due to the computational burden of matrix operations on microcontrollers, leading to slow execution. Other literature [[10], [11], [12], [13]] has presented low-cost PV emulators utilizing Buck converters governed by Arduino controllers; nevertheless, these systems suffer from low control performance and are limited to low-power regimes of only a few tens of Watts. Table 1 presents a comparison of recent studies reporting experimental results on the design of PV simulators. This comparison focuses on the proposed methods, along with their detailed advantages and disadvantages.Table 1Comparison of recent studies on PV simulator designs.Table 1 dummy alt textRef No.MethodAdvantagesDisadvantagesPublication year[14]PV emulation using the PHIL method with a lookup-table-based PV model, utilizing a dSPACE 1103 real-time simulator and an EXTECH 382,280 programmable power supply.- Simple lookup-table modeling.- Fast computation.- No closed-loop control; uses a low-pass filter with high latency.- Slow settling time (1.2 to 1.8 s).- Steady-state error of 2%.- Incapable of emulating partial shading conditions.2026[15]PV emulation using the PHIL method with a mathematical PV model, utilizing a dSPACE 1104 real-time simulator and a 220 W Buck converter.- High accuracy based on the mathematical PV model.- Complex model computation leading to a large calculation time (Ts=1ms).- Slow settling time (0.9 s).- Steady-state error of 2%.- Incapable of emulating partial shading conditions.2026[16]PV emulation using a simplified equivalent model based on a controlled current source combined with a diode and resistor.- Simple configuration directly using a current source, resistor, and diode.- Fast response with a settling time of 22μs.- Fails to accurately emulate the desired PV characteristics.- High thermal losses on the diode and resistor.- Incapable of emulating partial shading conditions.2023[17]PV emulation using a synchronous buck converter and a DSP F28335 (220 W capacity) using an explicit exponential model to reduce computational load.- Fast response with a settling time of 70 ms.- Capable of emulating partial shading conditions.- Steady-state error of 2.8%.- PV characteristic emulation error of approximately 3.08% compared to theory.2026[18]PV emulation using a synchronous buck converter and a DSP F28379D (220 W capacity) using an implicit mathematical model and a Newton–Raphson PV-model solver.- High accuracy according to the mathematical PV model.- Slow system response (∼15 s).- Long iterative model calculation time (∼200 μs).- Incapable of emulating partial shading conditions.2026This paperPV emulation using the PHIL method with a lookup-table-based PV model, utilizing an OPAL-RT real-time simulator and an APS1000 linear amplifier power supply.- High accuracy according to the mathematical PV model.- Extremely fast response speed (∼2 ms) based on closed-loop lookup-table control.- Capable of emulating partial shading conditions.- Requires a high-performance real-time simulator (OPAL-RT 5707XG).- Requires a PHIL power source with extremely low time delay (<5μs, APS1000 amplifier). Table 1Comparison of recent studies on PV simulator designs.Table 1 dummy alt textRef No.MethodAdvantagesDisadvantagesPublication year[14]PV emulation using the PHIL method with a lookup-table-based PV model, utilizing a dSPACE 1103 real-time simulator and an EXTECH 382,280 programmable power supply.- Simple lookup-table modeling.- Fast computation.- No closed-loop control; uses a low-pass filter with high latency.- Slow settling time (1.2 to 1.8 s).- Steady-state error of 2%.- Incapable of emulating partial shading conditions.2026[15]PV emulation using the PHIL method with a mathematical PV model, utilizing a dSPACE 1104 real-time simulator and a 220 W Buck converter.- High accuracy based on the mathematical PV model.- Complex model computation leading to a large calculation time (Ts=1ms).- Slow settling time (0.9 s).- Steady-state error of 2%.- Incapable of emulating partial shading conditions.2026[16]PV emulation using a simplified equivalent model based on a controlled current source combined with a diode and resistor.- Simple configuration directly using a current source, resistor, and diode.- Fast response with a settling time of 22μs.- Fails to accurately emulate the desired PV characteristics.- High thermal losses on the diode and resistor.- Incapable of emulating partial shading conditions.2023[17]PV emulation using a synchronous buck converter and a DSP F28335 (220 W capacity) using an explicit exponential model to reduce computational load.- Fast response with a settling time of 70 ms.- Capable of emulating partial shading conditions.- Steady-state error of 2.8%.- PV characteristic emulation error of approximately 3.08% compared to theory.2026[18]PV emulation using a synchronous buck converter and a DSP F28379D (220 W capacity) using an implicit mathematical model and a Newton–Raphson PV-model solver.- High accuracy according to the mathematical PV model.- Slow system response (∼15 s).- Long iterative model calculation time (∼200 μs).- Incapable of emulating partial shading conditions.2026This paperPV emulation using the PHIL method with a lookup-table-based PV model, utilizing an OPAL-RT real-time simulator and an APS1000 linear amplifier power supply.- High accuracy according to the mathematical PV model.- Extremely fast response speed (∼2 ms) based on closed-loop lookup-table control.- Capable of emulating partial shading conditions.- Requires a high-performance real-time simulator (OPAL-RT 5707XG).- Requires a PHIL power source with extremely low time delay (<5μs, APS1000 amplifier). Comparison of recent studies on PV simulator designs. Obviously, studies on PV emulator control based on accurate mathematical models require high computational load and cause significant control delays, as they typically solve mathematical models using iterative analytical methods such as Newton-Raphson and Gauss-Seidel. Equivalent approximation PV models, such as those employing the Lambert W function, still demand substantial computational resources, while Taylor series expansions remain accurate only within the neighborhood of the expansion point. Furthermore, achieving stable control with a fast response time for PV system emulation introduces significant difficulties in modeling and testing power converters involving numerous design parameters. To overcome these limitations, this study focuses on designing a simple and rapid PV power emulator system based on a lookup table with highly accurate modeled data. The proposed framework utilizes a real-time simulator to govern PV characteristics via a power amplifier, operating within a Power Hardware-in-the-Loop context. The core methodology of the proposed approach focuses on the following aspects:•Data modeling of PV modules under uniform temperature and irradiance profiles.•Data modeling of series-connected PV cells under non-uniform irradiance conditions.•Modeling the emulation system based on the parameters of the power amplifier and proposing a closed-loop controller combined with phase-lag compensation.•Development of a real-time PV characteristic controller based on look-up data tables. •Data modeling of PV modules under uniform temperature and irradiance profiles. Data modeling of PV modules under uniform temperature and irradiance profiles. •Data modeling of series-connected PV cells under non-uniform irradiance conditions. Data modeling of series-connected PV cells under non-uniform irradiance conditions. •Modeling the emulation system based on the parameters of the power amplifier and proposing a closed-loop controller combined with phase-lag compensation. Modeling the emulation system based on the parameters of the power amplifier and proposing a closed-loop controller combined with phase-lag compensation. •Development of a real-time PV characteristic controller based on look-up data tables. Development of a real-time PV characteristic controller based on look-up data tables. Method details In this study, we propose an accurate power emulation methodology for replicating photovoltaic (PV) characteristics based on Power Hardware-in-the-Loop (PHIL) simulation [19]. Unlike conventional methods that require designing and controlling complex power electronics circuits—which often suffer from challenging closed-loop I-V characteristic control and hardware design constraints—the proposed framework focuses on developing a high-accuracy descriptive PV model. This model is executed on a real-time simulator to govern a commercial power amplifier, precisely emulating the power characteristics of solar arrays under diverse environmental profiles without the necessity of developing complex customized controllers and power stages. In this research, a commercial PV module (Model: A10J-S72–175, manufactured by A10Green Technology) is employed for modeling and experimental evaluation. Its nominal technical specifications are summarized in Table 2.Table 2Technical specifications of the A10Green Technology A10J-S72–175 PV module.Table 2 dummy alt textParametersSymbolValuesMax power at STC (W)Pmpp175Voltage at Max Power (V)Vmpp36.63Current at Max Power (A)Impp4.78Open Circuit Voltage (V)Voc43.99Short Circuit Current (A)Isc5.17Nominal Operating Cell Temp ( °C)NOCT49.9Temperature coefficient of ISC (% / °C)ki0.042Temperature coefficient of VOC (% / °C)kv−0.362Temperature coefficient of Pmax (% / °C)kp−0.507 Table 2Technical specifications of the A10Green Technology A10J-S72–175 PV module.Table 2 dummy alt textParametersSymbolValuesMax power at STC (W)Pmpp175Voltage at Max Power (V)Vmpp36.63Current at Max Power (A)Impp4.78Open Circuit Voltage (V)Voc43.99Short Circuit Current (A)Isc5.17Nominal Operating Cell Temp ( °C)NOCT49.9Temperature coefficient of ISC (% / °C)ki0.042Temperature coefficient of VOC (% / °C)kv−0.362Temperature coefficient of Pmax (% / °C)kp−0.507 Technical specifications of the A10Green Technology A10J-S72–175 PV module. The implementation procedure of the proposed practical photovoltaic (PV) system emulation methodology consists of the following steps:•Step 1: High-accuracy data modeling of the PV system characteristics under two common scenarios: uniform environmental conditions (constant temperature and irradiance) and mismatched conditions under partial shading (non-uniform irradiance).•Step 2: Exporting the computed current-voltage (I-V) characteristic data arrays of the PV system from the model into the power emulation program within the MATLAB/Simulink environment.•Step 3: Designing a real-time PV emulator controller based on look-up data tables to minimize the computational burden and control execution time.•Step 4: Interfacing the power amplifier with the real-time emulator controller and the physical sensor circuits.•Step 5: Conducting experimental power emulation of the PV profiles under various simulated scenarios predefined by the calculated data tables. •Step 1: High-accuracy data modeling of the PV system characteristics under two common scenarios: uniform environmental conditions (constant temperature and irradiance) and mismatched conditions under partial shading (non-uniform irradiance). Step 1: High-accuracy data modeling of the PV system characteristics under two common scenarios: uniform environmental conditions (constant temperature and irradiance) and mismatched conditions under partial shading (non-uniform irradiance). •Step 2: Exporting the computed current-voltage (I-V) characteristic data arrays of the PV system from the model into the power emulation program within the MATLAB/Simulink environment. Step 2: Exporting the computed current-voltage (I-V) characteristic data arrays of the PV system from the model into the power emulation program within the MATLAB/Simulink environment. •Step 3: Designing a real-time PV emulator controller based on look-up data tables to minimize the computational burden and control execution time. Step 3: Designing a real-time PV emulator controller based on look-up data tables to minimize the computational burden and control execution time. •Step 4: Interfacing the power amplifier with the real-time emulator controller and the physical sensor circuits. Step 4: Interfacing the power amplifier with the real-time emulator controller and the physical sensor circuits. •Step 5: Conducting experimental power emulation of the PV profiles under various simulated scenarios predefined by the calculated data tables. Step 5: Conducting experimental power emulation of the PV profiles under various simulated scenarios predefined by the calculated data tables. Data modeling methodology for a photovoltaic (PV) module operating under uniform irradiance and temperature profiles The generalized single-diode model of a solar cell is illustrated in Fig. 1, where Rs represents the parasitic series resistance, Rsh denotes the parasitic shunt resistance, Iph signifies the photocurrent that varies linearly with the solar irradiance G, commonly referred to as the short-circuit current (Iph≈Isc), and Io is the diode reverse saturation current.Fig. 1Equivalent circuit of the single-diode solar cell model.Fig 1 dummy alt text Equivalent circuit of the single-diode solar cell model. The solar cell current characteristic is a non-linear function of voltage, as expressed in (1), whereas the photocurrent depends on temperature and solar irradiance according to (2), where ki represents the temperature coefficient of the short-circuit current [10], which is typically provided in the manufacturer's datasheet specifications.(1)I=Iph−Io[eV+I·Rsa1]−V+I·RsRsh(2)Iph=Isc.(1+ki.ΔT)G1000 The series resistance Rs and shunt resistance Rsh are calculated according to (3) and (4), respectively.(3)Rs=a·ln⁡(1ImppIph)+Voc−VmppImpp(4)Rsh=Vmpp·(Vmpp+Impp·Rs)Vmpp·Iph−Pmpp Where a represents the thermal voltage constant of the PV module, which determines the slope of the I-V curve in the vicinity of the Maximum Power Point (MPP) and depends on the cell temperature, the number of series-connected cells, and the diode ideality factor η. Typically, η ranges from 1.0 to 1.5 for conventional silicon, with a standard value of 1.2 frequently selected. Neglecting Rsh, the parameter a is calculated according to (5).(5)a=η·Ncell·Vt=η·Ncell·k·Trefq To ensure precise calculation and minimize modeling errors at the Maximum Power Point (MPP) relative to the manufacturer's nominal specifications, the reference parameter aref must account for the influence of Rsh and be calculated under each specific operating condition according to (6).(6)aref=Vmpp+Impp·Rs−Vocln(Iph−Impp−VmppRshIph−VocRsh) At this stage, the operating voltage of the solar cell can be precisely calculated for a given cell current and environmental condition according to (7), utilizing the Lambert W function [20,21] to simplify the solution of the transcendental equation (1).(7)V(I)=(Iph+Io−I)·Rsh−I·Rs−aref·W0(θ) Where the diode reverse saturation current [22,23] can be calculated according to (8), and the variable θ is determined by (9).(8)Io=Isc−VocRsheVocaref−1≈(IscVocRsh)·e−Vocaref(9)θ=Rsh·Ioaref·e(Iph+Io−I)·Rsharef We developed a MATLAB program to compute the photovoltaic (PV) current and voltage values based on the described mathematical model, which has been made publicly available on GitHub for readers' reference, cross-checking, and experimentation. The data modeling results of the PV module operating at a constant temperature of 25 °C under various irradiance levels [200, 400, 600, 800, 1000W/m2] are illustrated in Fig. 2.Fig. 2PV module data modeling results at a constant temperature of 25 °C under various irradiance levels [200, 400, 600, 800, 1000W/m2].Fig 2 dummy alt text PV module data modeling results at a constant temperature of 25 °C under various irradiance levels [200, 400, 600, 800, 1000W/m2]. To evaluate the accuracy and the goodness-of-fit between the theoretical mathematical PV model and the manufacturer's experimental data, the coefficient of determination R2 is utilized as expressed in (10). This metric is calculated based on two key quantities:•RSS (Residual Sum of Squares): Measures the deviation between the actual data points from the manufacturer's datasheet and the corresponding computed points derived from the PV module model.•TSS (Total Sum of Squares): Measures the total variation of the actual data points around their mean value.(10)R2=1−RSSTSS=1−∑k=1n(yk,datasheet−yk,model)2∑k=1n(yk,datasheet−y¯datasheet)2 •RSS (Residual Sum of Squares): Measures the deviation between the actual data points from the manufacturer's datasheet and the corresponding computed points derived from the PV module model. RSS (Residual Sum of Squares): Measures the deviation between the actual data points from the manufacturer's datasheet and the corresponding computed points derived from the PV module model. •TSS (Total Sum of Squares): Measures the total variation of the actual data points around their mean value.(10)R2=1−RSSTSS=1−∑k=1n(yk,datasheet−yk,model)2∑k=1n(yk,datasheet−y¯datasheet)2 TSS (Total Sum of Squares): Measures the total variation of the actual data points around their mean value.(10)R2=1−RSSTSS=1−∑k=1n(yk,datasheet−yk,model)2∑k=1n(yk,datasheet−y¯datasheet)2 In this evaluation, the goodness-of-fit is validated at three critical points provided in the manufacturer's datasheet for the A10Green Technology A10J-S72–175 PV module under Standard Test Conditions (STC):•Short-circuit point: (V=0,I=Isc)•Maximum Power Point (MPP): (V=Vmpp,I=Impp)•Open-circuit point: (V=Voc,I=0) •Short-circuit point: (V=0,I=Isc) Short-circuit point: (V=0,I=Isc) •Maximum Power Point (MPP): (V=Vmpp,I=Impp) Maximum Power Point (MPP): (V=Vmpp,I=Impp) •Open-circuit point: (V=Voc,I=0) Open-circuit point: (V=Voc,I=0) The calculated coefficient of determination results are R2 = 0.999,988 for current (I), R2 = 0.999,835 for voltage (V), and R2 = 0.999,998 for power (P). These values demonstrate an exceptionally high accuracy and strong correlation between the developed PV module model and the manufacturer's nominal specifications. Data modeling methodology for a photovoltaic (PV) module operating under non-uniform irradiance and temperature profiles In practical applications, PV cells as well as PV modules are commonly connected in series. Under non-uniform irradiance conditions—such as partial shading—the photocurrent Iph of individual cells is significantly affected. Consequently, the operating current may be forced to flow through the parasitic resistances (Rsh and Rs) of the shaded cells, leading to substantial power losses within the system [24]. Bypass diodes offer an effective solution to mitigate this issue by providing a low-impedance path that diverts the current away from Rsh and Rs. Fig. 3 illustrates a bypass diode integrated with a solar cell to bypass the reverse current flowing through the parasitic resistances.Fig. 3Equivalent circuit of the solar cell model with an integrated bypass diode.Fig 3 dummy alt text Equivalent circuit of the solar cell model with an integrated bypass diode. Under the STC with an irradiance of Gref=1000W/m2 and a temperature of 25 °C, as the operating temperature and solar irradiance decrease, the photocurrentIph,k, the open-circuit voltage Voc,k, and the diode reverse saturation current Io,kof the PV cell are calculated according to (11), (12), and (13), respectively.(11)Iph,k(Gk,T)=Iph·[1+ki·(T−Tref)]·(GkGref)(12)Voc,k(Gk,T)=Voc.[1+kv·(T−Tref)](13)Io,k=Iph,kexp⁡(Voc,kak)−1 Where the thermal voltage constant of the solar cell is approximated by neglecting Rsh as expressed in (14), while the temperature-dependent series resistance Rs is approximated according to (15).(14)ak=Vmpp,k+Impp,k·Rs−Voc,kln(1Impp,kIsc,k)(15)Rs=Rs_ref·(TTref) To simplify the calculation of the solar cell voltage, the influence of Rsh is neglected, allowing Eq. (7) to be straightforwardly solved through algebraic transformation according to (16).(16)Vmod,k=ak·ln(Iph,k−IstringIo,k+1)−Rs·Istring Fig. 4 considers three cell-groups in the A10Green Technology A10J-S72–175 72-cell PV modules, with each group consisting of 24 cells connected with a bypass diode subjected to non-uniform irradiance levels of 1000, 600, and 300 W/m2, respectively.Fig. 4Modeling of a PV module with 72 series-connected cells under non-uniform irradiance.Fig 4 dummy alt text Modeling of a PV module with 72 series-connected cells under non-uniform irradiance. The calculation of the PV module voltage within a under non-uniform irradiance is evaluated based on the following two cases:Case 1: The module current exceeds the photocurrent of the kth PV cell group (Imodule≥Iph,k) At this stage, the module voltage equals the forward bias voltage of the bypass diode, as the current flows through the diode.(17)Vgroup,k=−VbypassCase 2: The PV cell-group current is lower than the photocurrent of the PV module (Imodule6dB). The system can withstand load variations of up to +7.42dB without losing stability. The first notch frequencies of the moving average filter (MAF) effectively eliminate high-frequency noise regions or typical PWM switching oscillations. The plot clearly illustrates infinitely deep attenuation points (attenuation of -60dB) precisely at the PWM carrier frequency multiples (kfsw), effectively filtering out current switching harmonic ripples before feeding back to the controller.Fig. 7System stability evaluation using Bode and Nyquist diagrams.Fig 7 dummy alt text System stability evaluation using Bode and Nyquist diagrams. In addition, Fig. 7(b) illustrates the Nyquist trajectory of the open-loop system L(s), which does not encircle the critical point (−1,j0). Since the open-loop system has no poles in the right-half complex plane, according to the Nyquist stability criterion, the closed-loop system is absolutely stable. The intersection of the Nyquist trajectory with the negative real axis occurs at the point ≈−0.426. The distance from this point to the critical point (−1,j0) represents the gain margin GM=20log10(1/0.426)≈7.42dB. The design ensures that the system does not experience resonant oscillations even when the load parameters R vary. To evaluate the closed-loop control system response, we construct a unity negative feedback system with a filter in the feedback path, and the closed-loop transfer function is given by (30).(30)Gclosed(s)=I(s)I*(s)=Gfw(s)1+Gfw(s)·Hfeedback(s)=Gfw(s)1+Gopen(s) Where:•Gfw(s)=Gctrl(s)·Gplant(s) is forward transfer function.•Hfeedback(s) is the feedback path transfer function.•Gopen(s)=Gfw(s)·Hfeedback(s) is the overall open-loop transfer function. •Gfw(s)=Gctrl(s)·Gplant(s) is forward transfer function. Gfw(s)=Gctrl(s)·Gplant(s) is forward transfer function. •Hfeedback(s) is the feedback path transfer function. Hfeedback(s) is the feedback path transfer function. •Gopen(s)=Gfw(s)·Hfeedback(s) is the overall open-loop transfer function. Gopen(s)=Gfw(s)·Hfeedback(s) is the overall open-loop transfer function. To calculate the exact analytical closed-loop transfer function, we substitute Gfw(s) and Gopen(s) calculated in the previous steps into the general formula (30), yielding Gclosed(s) as follows:Gclosed(s)=1.0124·(s+5000s)·(1+1.509×10−5s1+2.623×10−5s)e−s·5×10−61+20248s2(s+50001+2.623×10−5s)(1+1.509×10−5s)(1−e−s·50×10−6)e−s·11×10−6 Applying the first-order Padé approximation yields the polynomial-approximated closed-loop transfer function as follows:Gclosed,approx(s)≈1.0124·(s+5000)·(1+1.509×10−5s)·(1+1.8×10−5s)·(1−2.5×10−6s)s·(1+2.623×10−5s)·(1+1.8×10−5s)+1.0124·(s+5000)·(1+1.509×10−5s)·(1−1.8×10−5s) Fig. 8 presents the model and closed-loop control response results via the delay compensation and PI stages. The results in Fig. 8(b) show that the overshoot is 2.98% thanks to the very large phase margin (PM=86.91∘) provided by the Lead-PI compensator. Specifically, the current response speed increases from Iinitial=0A to the peakIpeak=5.15A in an extremely short rise time tr≈30μs. This speed accurately reflects the high open-loop bandwidth (fc=5.41kHz). The minor ripple after the first peak is due to the characteristic effect of the total time delay Td=36μs and the MAF filter (with notch points at k×fsw). However, these ripples are damped very quickly, and the current settles smoothly to its steady-state value.Fig. 8Evaluation of closed-loop current control response via PI delay compensation.Fig 8 dummy alt text Evaluation of closed-loop current control response via PI delay compensation. Thus, the Lead-PI + MAF controller completely solves the hardware delay problem Td=36μs, effectively rejects noise while maintaining extremely fast current tracking, negligible overshoot, and stable operation. Method validation To validate the proposed PV system emulation method, an experimental hardware setup was constructed, consisting of an OP5707XG real-time simulator, a 1 kW APS1000 controllable power amplifier, an LA25-P current sensor with a maximum measurement range of 25 A, and a 100Ω power rheostat for testing. Fig. 9 illustrates the experimental setup of the proposed PV emulator utilized for the validation in this paper.Fig. 9Experimental setup of the proposed PV emulator.Fig 9 dummy alt text Experimental setup of the proposed PV emulator. In this experimental setup, the emulated PV current signal is acquired from the LA25-P current sensor and fed into the OP5340 analog input card, which features a sampling rate of up to 500 kS/s for measuring ±20 V input signals. Concurrently, the control voltage signal for the PV emulator is routed from the OP5330 analog output card—capable of simultaneous sampling at 1 MS/s with a voltage range of ±16 V to the input of the APS1000 power amplifier, which exhibits a slew rate of up to 52 V/us at no-load, signal delay between amplifier’s external input and amplifier’s output < 5 µs, and a full-load response time of 2–3 ms according to the manufacturer's specifications. Fig. 10 illustrates the schematic of the signal connections and the power wiring diagram of the experimental setup.Fig. 10Connection schematic of the proposed PHIL-based PV emulator using OPAL-RT.Fig 10 dummy alt text Connection schematic of the proposed PHIL-based PV emulator using OPAL-RT. Fig. 11 illustrates the actual signal connections and power wiring of the experimental setup. In this experiment, three turns of wire were wound through the primary aperture of the LA25-P current sensor to increase the primary current ratio, thereby matching the sensor's measurement range with the 5 A short-circuit current of the PV module under study.Fig. 11Hardware connections between the OPAL-RT OP5707 real-time simulator, the APS1000 power amplifier, and the sensor board.Fig 11 dummy alt text Hardware connections between the OPAL-RT OP5707 real-time simulator, the APS1000 power amplifier, and the sensor board. The OPAL-RT real-time simulator executes the computational control program compiled from MATLAB/Simulink. The connection configuration with the analog signal interface cards is managed within the RT-LAB software. Fig. 12 illustrates the physical hardware connection configuration in RT-LAB via the OP5340 analog input card and the OP5330 analog output card.Fig. 12Hardware configuration of the OP5707 interfacing with the sensor and power amplifier within the RT-LAB software.Fig 12 dummy alt text Hardware configuration of the OP5707 interfacing with the sensor and power amplifier within the RT-LAB software. Case study 1: measurement of PV emulation responses at 25 °C under varying irradiance levels By adjusting the rheostat from 0 to 100Ω across five experimental runs—corresponding to the PV emulation utilizing five irradiance lookup tables of [0.2, 0.4, 0.6, 0.8, 1.0] kW/m2—the measured characteristic curves of the A10Green Technology A10J-S72–175 PV module were obtained from the monitoring screen of the APS1000 power amplifier, as illustrated in Fig. 13. The evaluation results of the coefficient of determination R2 between the theoretical model and the experimental measured data at various irradiance levels show that in the high-irradiation region (1000W/m2 to 600W/m2), the R2 value remains stable at an extremely high level (R2≥0.9997), indicating that the simulation model accurately reflects the physical phenomena of the photovoltaic panel under strong light. However, in the low-irradiation region (200W/m2), the R2 coefficient decreases slightly to 0.9951 because the short-circuit current (Isc) and the output power of the PV panel are very small (a 5-fold reduction compared to standard test conditions). At this point, the signal-to-noise ratio (SNR) of the measurement sensors, data acquisition (DAQ) circuitry, and communication cards decreases, resulting in relatively larger errors compared to the high-irradiation region. The experimental results demonstrate a good agreement between the actual operational points under different irradiance levels and the theoretical curves. However, near the short-circuit point, voltage and current oscillations are observed due to the inherent nature of the voltage-source power amplifier.Fig. 13Experimental measurement results of the PV emulator at 25 °C under various irradiance levels of [0.2 0.4 0.6 0.8 1] kW/m2.Fig 13 dummy alt text Experimental measurement results of the PV emulator at 25 °C under various irradiance levels of [0.2 0.4 0.6 0.8 1] kW/m2. Fig. 14 illustrates the response of the proposed PV emulator system when a sudden load step is applied from no-load to full-load with the PV panel operating under STC, with a designed proportional gain value of Kp=101. The results show that under no-load control, the current is 0A shown in CH2 and the open-circuit voltage is approximately 44V shown in CH3. When the test resistive load is connected, the voltage rapidly reaches 37.084V and the current reaches 4.715A, with a switching settling time of about 2ms and a current overshoot of 27% at the rated power of 175W. The results reveal a significant discrepancy between the closed-loop transient response and the experimental response. This is largely due to the system delay modeling, in which we adopted the maximum delay time specified by the amplifier and sensor manufacturers.Fig. 14Response of the proposed PV emulator system under sudden full load changes with Kp = 101.Fig 14 dummy alt text Response of the proposed PV emulator system under sudden full load changes with Kp = 101. To reduce the current overshoot in the closed-loop control on the hardware prototype, we reduced the PI controller gain from the initial design value of Kp=101 to an adjusted value of Kp=71. The resulting simulated current and voltage responses of the PV panel are illustrated in Fig. 15. As can be seen, the current overshoot decreases to approximately 4% after adjusting the gain. Furthermore, the steady-state error is approximately zero, indicating that the proposed controller exhibits good dynamic response suitable for high-precision PV emulation research with static loads such as resistors.Fig. 15Response of the proposed PV emulator system under sudden full load changes with Kp = 71.Fig 15 dummy alt text Response of the proposed PV emulator system under sudden full load changes with Kp = 71. To evaluate the response of the proposed PV emulator for dynamic load components featuring high-frequency pulse switching, a commercial Suoer SON-MPPT-20A solar charger was used for testing. Fig. 16(a) illustrates the experimental setup of the emulator on an MPPT solar charging circuit, featuring a battery pack consisting of two 12V/40Ah batteries connected in parallel with the charger. The solar charger integrates a 3-stage charging function, a Maximum Power Point Tracking (MPPT) function within a voltage range of 18–150 V, and automatic battery voltage recognition.Fig. 16Experimental testing of the PV emulator on a commercial solar MPPT charger.Fig 16 dummy alt text Experimental testing of the PV emulator on a commercial solar MPPT charger. Fig. 16(b) depicts the panel voltage and current responses during the test. When the emulator operates, the open-circuit voltage rapidly reaches the panel's open-circuit value of 44 V. Once the panel voltage reaches 44 V, the charger initiates a 5-second check of the photovoltaic and battery voltage states. Subsequently, the charger sweeps the panel operating voltage from the open-circuit voltage of 44 V down to 16 V within 6 s. Upon completing the panel voltage sweep, the charger activates continuous MPPT control at the maximum power point, with the tracked voltage fluctuating between 35 V and 38 V. The test results on the commercial charger show that the voltage and current control responses of the emulator are stable in the region where the PV panel power reaches its peak of 175 W. However, in the voltage region below 30 V, the current exhibits significant fluctuations. The core cause of these oscillations stems from the PV panel's characteristic of extremely low current variation (ΔI≈0.12A) across voltages from 0 V to 30 V (corresponding to currents from 5.17A to 5.05A). Furthermore, the high-frequency operation of the charger with complex dynamic characteristics, along with significantly increased noise spikes from the current and voltage sensors during operation, contributes to higher current oscillations compared to resistive loads. Case study 2: measurement of PV emulation responses under partial shading conditions with three cell-groups at different irradiance levels The simulation lookup tables for a PV module of three cell-groups were modified in MATLAB/Simulink and subsequently compiled within the RT-LAB software. By adjusting the rheostat from 0 to 100Ω across five experimental runs—corresponding to temperature variations of [25, 35, 45, 55, 65] °C for emulating the three cell-groups PV module under respective irradiances of 300, 600, and 1000 W/m2. The measured characteristic curves of the three cell-groups series-connected A10Green Technology A10J-S72–175 PV modules were obtained, as illustrated in Fig. 17. The experimental results demonstrate that the actual operational points of the PV module under different temperature conditions show good agreement with the theoretical curves.Fig. 17Experimental measurement results of the of three cell-groups PV module emulation under individual module irradiances of 300, 600, and 1000 W/m2 across temperature cases of [25, 35, 45, 55, 65] °C.Fig 17 dummy alt text Experimental measurement results of the of three cell-groups PV module emulation under individual module irradiances of 300, 600, and 1000 W/m2 across temperature cases of [25, 35, 45, 55, 65] °C. Fig. 18 illustrates the power characteristics of the three series-connected PV modules obtained from the MATLAB/Simulink simulation lookup tables, alongside the actual power operational points measured from the experimental setup. The evaluation results of the coefficient of determination R2 between the theoretical model and the experimental measured data at temperature levels from 25∘C to 65∘C are all approximately 1.0 (ranging from 0.9997 to 1.0000). This proves that the mathematical model simulating the I-V and P-V characteristics of the PV Emulator system closely tracks the actual data collected over the entire investigated temperature range without significant degradation as the temperature increases.Fig. 18Measured power curves of the of three cell-groups PV module emulation under individual irradiances of [300, 600, 1000] W/m2 and various temperature conditions.Fig 18 dummy alt text Measured power curves of the of three cell-groups PV module emulation under individual irradiances of [300, 600, 1000] W/m2 and various temperature conditions. We utilized the partial shading characteristics of the PV panel at 25∘C, with irradiance levels for the three cell groups set to 300, 600, and 1000 w/m2, respectively, to evaluate the shaded panel simulation on the solar MPPT charger. Fig. 19 illustrates the voltage and current responses of the PV panel under the aforementioned shading condition. It can be observed that during the 6-second sweep process, the voltage decreases from an open-circuit voltage of 44V down to 16V. During this process, current appears in two distinct irradiance regions. Upon completion of the sweep, the charger activates the MPPT control to operate at the Global Maximum Power Point (GMPP) around 24V with a maximum power of 68W. The results demonstrate that the proposed emulator can be effectively used to evaluate MPPT charger controllers under standard partial shading conditions.Fig. 19Experimental results of the PV emulator on the solar MPPT charger under shaded conditions.Fig 19 dummy alt text Experimental results of the PV emulator on the solar MPPT charger under shaded conditions. In summary, all experimental results demonstrate that the proposed PV emulation method delivers an accurate response, making it highly suitable for evaluating PV characteristic behaviors on a PV emulator platform. Limitations The emulation experimental results may vary if a power amplifier with a slower response time is utilized, or when incorporating electronic loads—such as solar inverters and solar chargers—that feature high-frequency PWM switching control. In addition, the OPAL-RT real-time simulator does not support PHIL simulation for mathematical models based on numerical iteration methods; thus, model-based testing cannot be implemented, which makes continuously varying irradiance and temperature require multiple look-up tables and complex interpolation. Furthermore, the capability to configure and interpolate PV characteristics directly based on standard Solar Array Simulator (SAS) models, such as EN50530 or Sandia (as found in commercial emulators), has not yet been implemented. CRediT author statement Hoai Phong Nguyen: Conceptualization, Validity tests, Data curation, Methodology, Software, Writing- Original draft preparation, Thuan Thanh Nguyen: Conceptualization, Visualization, Investigation, Supervision, Reviewing and Editing. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.