Current Mode Control of a Solar Inverter with MPPT Algorithm

Izsák Ferencz, Dorin Petreuș

Department of Applied Electronics
Technical University of Cluj-Napoca, Romania
[email protected]

Abstract: In the last decade, grid-connected photovoltaic systems were studied and developed due to
advance in the technology of solar panels, their price drop and the climatic crisis. This work will study,
analyze and design a low power solar inverter. The most important design constraints for a solar
inverter are: efficiency, maximum power point tracking (MPPT) performance and synchronization with
the utility grid. The objective of this project is to find a solution for each (minimization of power losses,
precise tracking of grid frequency). To achieve a satisfying efficiency, a current mode control loop in a
two-dimensional control frame was implemented, with an enhanced phase-lock loop (EPLL) as the
synchronization mechanism with the utility grid. For tracking the maximum point on the power curve
of the solar cell, a classical hill climbing method was improved and integrated into the system. For
simulations, Matlab Simulink software was used and input power versus output power was examined
along with total harmonic distortion (THD) of the inverter’s output current, while for experimental
testing a dSPACE controller board and two Suntec Power solar panel. The paper is focused on the
current mode control loop mathematical equations providing a simplified step by step design process.

1. INTRODUCTION adaptive filter enhanced PLL was designed to obtain

adaptive noise cancelling (ANC) in its control. [8,9]
Solar power is an infinite energy source with great
potentials and its applications has shown significant In [1-4] different control algorithms are presented
growth during the last decade. Among other in the DQ control frame with problems appearing like
renewables, solar energy has shown a rapid evolution oscillations or poor tracking ability at large jumps of
in technology and accessibility, becoming much irradiance, which means a slow response. This work
cheaper. To integrate it into the utility grid, a proper presents a simple current mode control with good
solar inverter system must be designed. The goal of tracking ability and easy design, capable to follow
this project is to choose and optimize methods in order variations fast and precise.
to obtain a fast and reliable system. The designed solar inverter topology has the
In this project a single-phase inverter was used; a advantages of a simple and efficient control, reliable
smooth sinusoidal output was obtained with high- grid synchronization and good power transfer ratio. It
speed power semiconductors so that the harmonic also provides a simplified design process as and it can
contents at the output to be reduced by pulse width be easily rescaled for higher powers.
modulation techniques. [6] In a voltage source inverter The paper is structured as follows: section 2
(VSI) system, the sinusoidal voltage and current presents the system description, section 3 presents the
commands must be tracked, which is a difficult current mode control principle, section 4 the
compensator design task. The αβ-frame and dq-frame simulations and results and section 5 the conclusions.
are two main two-dimensional frames, which
simplifies the problem of controlling the inverter and 2. SYSTEM DESCRIPTION
makes possible power control also. In the dq-frame, in
steady state, the signals are assumed to be DC Fig. 1 presents the whole system schematic of the
waveforms, which permit the usage of simpler solar inverter connected to the grid. The modelled
structure and lower dynamic compensators. [7] To solar panel is two Suntech Power STP075S-12/Bb
phase-lock the internal reference with the grid, an

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 1 2019 42nd International Spring Seminar on Electronics Technology (ISSE)
 Fig. 1. The whole system in Matlab Simulink.

connected in series. A single phase full-bridge inverter the MPP at different irradiance values of the used solar
is then used to obtain the DC energy from the panel panel. In the case of incremental conductance method
and convert it into AC energy, transferring it to the the change of voltage is tested, but the MPP is more
grid. The inverter is synchronized with the grid and closely tracked by the current and power, because the
the control loops using the enhanced PLL (EPLL) current extracted from the PV panel is directly
block. Below them, the control loop can be seen, proportional with the irradiation. Therefore, the slope
consisting of the main current loop transposed in the of PV current and power is similar to the irradiation’s:
dq rotating frame with two axis, one for active power the power changes in the same direction as current
and one for reactive power. The MPP tracking (see Fig. 6). The solution can be testing the change of
(MPPT) algorithm block provides the reference for the current instead of the voltage and the usage of ΔI/ΔP
active power. instead of just ΔI, when ΔI is negative. The obtained
algorithm can be observed in Fig. 3. [5,10,11]
2.1. Inverter design
The full-bridge converter is composed of four
switch cells, two inverter branches with upper and
lower cells. Each of these switch cells are realized by
antiparallel connection of a fully controllable
unidirectional IGBT and a known reverse-conducting
diode switch. From the DC side the converter is fed by
two equal voltage sources V/2, in this case the two
capacitors charged with the voltage delivered by the
solar panel, whose common point is node zero – a DC
side midpoint chosen to be a voltage reference node.
The converter’s operation is based on alternate
switching of the transistor pairs. The turn-on/off
command of the transistors is made through pulse-
width modulation (PWM) techniques. [6]
2.2. MPPT algorithm
For the MPPT algorithm a classical hill climbing
method was chosen and improved. Fig. 2 illustrates Fig. 2. The output characteristics of the solar panel.

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 2 2019 42nd International Spring Seminar on Electronics Technology (ISSE)
 parameters and makes possible the control of the real
and reactive powers exchanged between the DC and
AC side, but the control variables and feedback signals
are still sinusoidal functions of time. Fortunately, the
dq-frame control also has all the benefits of the αβ-
frame control, furthermore the advantage of DC
quantities as control variables in steady state, which is
a great facility in compensator design. Since the
control variables are DC quantities, the zero steady-
state error can be achieved simpler by adding an
integral component to the compensator structure [7].
A symmetrical three-phase system can be described
by a space phasor equation:
2
𝑓⃗ 𝑡 𝑒 𝑓 𝑡 𝑒 𝑓 𝑡 𝑒 𝑓 𝑡 1
Fig. 3. The modified incremental conductance algorithm. 3

2.3. The EPLL where fa,b,c are sinusoidal functions with amplitude and
different initial phase angles. Decomposing it using
Using adaptive filtering technique to cancel some 𝑓⃗ 𝑡 𝑓 𝑡 𝑗𝑓 𝑡 the αβ transform is derived as:
specific high frequency components, the filter is also
known as adaptive notch filter (ANF). [9] The 1 1
⎡1 ⎤ 𝑓 𝑡
𝑓 𝑡 2 2 2⎥
adaptive notch filter operates based on the least-mean- ⎢ 𝑓 𝑡 2
squares (LMS) algorithm. It estimates the gradient on 𝑓 𝑡 3⎢ √3 √3⎥
𝑓 𝑡
the mean-square error surface to find a proper weight ⎣0 2 2⎦
vector at the point where the mean-square error point By further phase-shifting the system by –ε(t) using
is minimum. [8] The ANC system is attached to a 𝑓 𝑗𝑓 𝑓 𝑗𝑓 𝑒 , the dq transform is obtained:
classic PLL structure, being applied to the phase-
𝑓 𝑡 cos 𝜀 𝑡 sin 𝜀 𝑡 𝑓 𝑡
detector multiplier, enhancing its performance. 3
𝑓 𝑡 sin 𝜀 𝑡 cos 𝜀 𝑡 𝑓 𝑡
In the dq reference system the form of the active
(P) and reactive (Q) power is:
3
P t V t i t V t i t
2 4
3
Q t V t i t V t i t
Fig. 4. The block diagram of the EPLL. 2
In Fig. 4 can be observed that in the moment when where Vsd and Vsq are dq-frame voltage components of
the frequency and the phase-angle of the voltage- the grid-connected AC system:
controlled oscillator (VCO) generated reference signal
x = sin(ωt) matches those of the input signal v, the 𝑉 𝑉 cos 𝜔 𝑡 𝜃 𝜌
5
output of the ANF will become zero. In this way at the 𝑉 𝑉 sin 𝜔 𝑡 𝜃 𝜌
output of the phase-detector (PD) multiplier the signal
In equation (4) it can be observed that the powers
oscillations are eliminated, the signal phase-angle is
are dependent of id and iq currents. The role of the PLL
detected properly by the EPLL and then provided for
is to maintain vsq zero, so that the two powers could be
the αβ-dq blocks. To track smoother the frequency
controlled by id and iq respectively as follows [6]:
variations, low-pass filter was also added. [8]
3
𝑃 𝑡 𝑉 𝑡 𝑖 𝑡
2
3. CURRENT MODE CONTROL IN DQ FRAME 3
6
𝑄 𝑡 𝑉 𝑡 𝑖 𝑡
2
The αβ-frame control of a VSC system with grid-
imposed frequency reduces the number of control

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 3 2019 42nd International Spring Seminar on Electronics Technology (ISSE)
 If the control loop tracks fast enough, the reference Using (10), two first order, decoupled linear system
id ≈ idref and iq ≈ iqref , then Ps ≈ Psref and Qs ≈ Qsref , is obtained, where id and iq are controlled by ud and uq
which means that Ps(t) and Qs(t) can be controlled respectively - the outputs of two compensators that
independently. Thus, if Psref and Qsref are constant, all process the error on each axis PI controllers, in this
the variables are DC variables [6]. case, ed = idref – id and eq = iqref – iq.
2 𝑑𝑖
⎧ 𝑖 𝑃 𝑡 𝐿 𝑅 𝑟 𝑖 𝑢
3𝑉 𝑑𝑡 11
7 𝑑𝑖
⎨𝑖 2
𝑄 𝑡 𝐿 𝑅 𝑟 𝑖 𝑢
⎩ 3𝑉 𝑑𝑡

Because zero reactive power is wanted, the Therefore, the loop gain is dependent of L and R.
reference for the Q power control branch (iqref) is set to This means that the proportional and integrative
zero. The reference signal for the P power however is constants of the PI can be deduced as a function of L
obtained from the MPPT block through a proportional and R as kp=L/τi and ki=(R+ron)/τi, τi being a design
integrative block to track variations of solar panel choice time constant.
current and voltage and obtain the maximum power in
each given condition. The id and iq currents are
obtained through the abc to αβ transfer block from the
inverter output current. Starting from the space-phasor
form the AC side dynamics of the VSI can be
deduced:
𝑑𝑖
⎧𝐿 𝐿𝜔 𝑡 𝑖 𝑅 𝑟 𝑖 𝑉 𝑉
⎪ 𝑑𝑡
𝑑𝑖
𝐿 𝐿𝜔 𝑡 𝑖 𝑅 𝑟 𝑖 𝑉 𝑉 8
⎨ 𝑑𝑡
⎪ 𝑑𝜌
⎩ 𝜔 𝑡
𝑑𝑡
Equation (8) shows that ρ(t) = ω0(t) + θ0 implies Fig. 5. The control diagram of the grid connected PV syst.
that Vsq = 0. Thus, the PLL is devised to regulate Vsq
at 0 by the feedback law 𝜔 𝑡 𝐻 𝑝 𝑉 𝑡 , where The obtained control signals are transformed back
H(p) is a linear transfer function of the PLL control to αβ-frame and fed to the PWM generator to send the
loop. Therefore, the PLL adjusts the rotational speed command pulses for the inverter. The EPLL provides
of the dq-frame, ω, in such way that Vsq to be zero in the proper rotational speed for the dq-frame, ωt, to be
steady state. Assuming steady-state operating synchronized with the grid frequency and phase. The
condition (ω(t) = ω0), then (8) can be written as: control system’s block diagram is presented in Fig. 5.
𝑑𝑖
𝐿 𝐿𝜔 𝑖 𝑅 𝑟 𝑖 𝑉 𝑉 4. SIMULATIONS AND RESULTS
𝑑𝑡 9
𝑑𝑖
𝐿 𝐿𝜔 𝑖 𝑅 𝑟 𝑖 𝑉 𝑉 The control algorithm and the simulations were
𝑑𝑡
implemented in Matlab Simulink tool, which is
compatible with different real platforms, in this
where 𝑉 𝑡 𝑚 𝑡 and 𝑉 𝑡 𝑚 𝑡 ,
experiment with a dSPACE DS1104 controller board.
representing the VSC model in dq-frame. Since the
presence of Lω0 terms id and iq are dynamically The current and power extracted from the panel are
coupled. To decouple the currents (eliminating Lω0), the quantities, which vary the most. The voltage
md and mq are determined as: remains constant enough to be a good reference,
therefore the MPPT algorithm is built over a small
2
⎧𝑚 𝑢 𝐿𝜔 𝑖 𝑉 voltage interval that is around the MPP of the panel’s
𝑉 P-V curve. This power-proportional characteristic (see
10
⎨𝑚 2 Fig. 6 below) of the inverter current is used in the
𝑢 𝐿𝜔 𝑖 𝑉
⎩ 𝑉

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 4 2019 42nd International Spring Seminar on Electronics Technology (ISSE)
current-mode control loop to obtain a maximum power
with the proper reference currents.

Fig. 8. D and Q axis compensator signals.

To simulate the derived EPLL system, a frequency
Fig. 6. Irradiance, panel current and power proportionality.
step from a higher value to a lower one was set. The
additional loop filter ensures a smooth ω waveform.
The MPPT algorithm block generetes a reference
voltage from which the panel voltage is subtracted to
feed a PI controller. The obtained signal is the
reference current id (iq is set to 0 as already discussed).

Fig. 9. EPLL with the additional Low-Pass filter with cutoff

frequency of 50Hz.
Fig. 7. Id and Iq signals In Fig. 9 can be observed that the EPLL systems
The inverter current is transformed into dq-frame provides an output with zero steady-state error.
quantities with the αβ to dq block to obtain the control As a final step, the power transfer is examined. On
“currents” Id and Iq (Fig. 7). The current mode control Fig. 10 one can notice the reactive power (grey) and
loop substracts these from the reference currents the active power (orange) with the DC panel power
mentioned above and a PI controller for each axis superposed (blue); the power loss is minimal even at
produces the control signals, see on Fig. 8 below: irradiance jumps.

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 5 2019 42nd International Spring Seminar on Electronics Technology (ISSE)
 Having the experimental setup from Fig. 12, to test
the system, the following signal waveforms were
obtained as seen in Fig. 13:

Fig. 10. Variation of reactive power as a function of

irradiance jump and the input-output active powers.
Fig. 13. The obtained waveforms on the oscilloscope, the
No harmonic correction was implemented in the inverter’s output current and voltage with the EPLL
control algorithm; therefore, the third harmonic has a provided synchronization signal.
strong presence almost as the international maximum
allowed value of 4%. Also, the total harmonic 5. CONCLUSIONS
distortion (THD) is near the required 5%. This can be
further improved with a 3rd harmonic filter or a The paper presented a simple current mode control
harmonic compensation loop. See the result on algorithm, its working principle, mathematical model
Fig. 11. and design parameters. Along with a good MPPT and
EPLL loop, the obtained results are according to the
theoretical demonstrations, showing that a single-
phase system can be optimized to minimal power
losses and a fast response, as required. Furthermore it
is a rescalable system, allowing the testing of different
MPPT algorithms.
This study shows the robustness of the dq
rotational reference frame in applications of solar
inverter control; moreover, the applicability of Matlab
Fig. 11. Total Harmonic Distortion (THD) of inverter output and dSPACE control board for rapid prototyping.
current at 1000W/m2.
REFERENCES
[1] A. M. Mnider, D. J. Atkinson, M. Dahidah, and M.
Armstrong, “A simplified DQ controller for single-
phase grid-connected PV inverters,” IREC 2016 - 7th
Int. Renew. Energy Congr., pp. 1–6, 2016.
[2] B. Li, S. Huang, X. Chen, and Y. Xiang, “A simplified
DQ-frame current controller for single-phase grid-
connected inverters with LCL filters,” 2017 20th Int.
Conf. Electr. Mach. Syst. ICEMS 2017, 2017.
[3] T. Zhao et al., “An optimized third harmonic
compensation strategy for single-phase cascaded H-
bridge photovoltaic inverter,” IEEE Trans. Ind.
Electron., vol. 65, no. 11, pp. 8635–8645, 2018.
Fig. 12. The experimental setup in the laboratory.
[4] M. Ebrahimi, S. A. Khajehoddin, and M. Karimi-
Ghartemani, “Fast and Robust Single-Phase DQ
Current Controller for Smart Inverter Applications,”

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 6 2019 42nd International Spring Seminar on Electronics Technology (ISSE)
 IEEE Trans. Power Electron., vol. 31, no. 5, pp. 3968– [8] Remus Teodorescu, Marco Liserre and Pedro
3976, 2016. Rodríguez, Grid converters for photovoltaic and wind
[5] Sumedha Sengar “Maximum Power Point Tracking power systems, Published by John Wiley & Sons, Ltd.,
Algorithms for Photovoltaic System: A Review 2011
“International Review of Applied Engineering [9] Simon Haykin, Adaptive filter theory, Prentice Hall, 3rd
Research ISSN 2248-9967 Volume 4, Number 2 edition 1995.
(2014), pp. 147-154 [10] Mihnea Rosu-Hamzescu, Sergiu Oprea, Microchip
[6] Amirnaser Yazdani, Reza Iravani, Voltage-Sourced Technology Inc. “Practical Guide to Implementing
Converters in Power Systems, Published by John Wiley Solar Panel MPPT Algorithms” 2013 Microchip
& Sons, Inc., Hoboken, New Jersey, 2010. Technology Inc.
[7] T. M. Rowan and R. J. Kerkman, “A New Synchronous [11] David Sanz Morales “Maximum Power Point Tracking
Current Regulator and an Analysis of Current- Algorithms for Photovoltaic Applications” M. Eng.
Regulated PWM Inverters,” IEEE Transactions on Thesis, Aalto University School of Science and
Industry Applications, vol. IA-22, no. 4, pp. 678–690, Technology, Espoo, Finland, 2010.
March/April 1986.

978-1-7281-1874-1/19/$31.00 ©2019 IEEE 7 2019 42nd International Spring Seminar on Electronics Technology (ISSE)