Академический Документы
Профессиональный Документы
Культура Документы
Submitted: 19.12.2018
I
Contents
1. OVERVIEW -1-
3. CONCEPT - 47 -
4. TESTS - 56 -
4.1. Simulations - 56 -
4.1.1. Simulation of the PV-panel - 56 -
4.1.1. Simulations of the TI-SEPIC in LTspice and Simulink - 59 -
4.1.2. Simulink model: SEPIC connected with the PV panel - 60 -
5. RESULTS - 81 -
6. DISCUSSION - 98 -
III
7. CONCLUSION - 100 -
IV
Notes
Photovoltaic, Power Converter, Maximum Power Point (Tracker), SEPIC Converter, Battery
Charging, Solar Charging, LTspice, MATLAB Simulink, Circuit Simulation, STM
Microcontroller, Texas Instruments Incorporated Solar Explorer Kit
Short Summary
The goal of this project is to use a small photovoltaic panel with a maximum power output of
50 Watts to charge a battery. The charging circuit must be developed, and the DC/DC
converter stage built. Different MPPT algorithm techniques shall be compared, a battery
storage chosen and an MPPT method implemented. A given Microcontroller must be used to
achieve this task. Besides that, the charging circuit shall be simulated and compared with
tests of the designed converter concept.
V
Index of Abbreviations
PV Photovoltaic
MPP Maximum Power Point
MPPT Maximum Power Point Tracker
CCM Continuous Current Mode
DC Direct Current
AC Alternating Current
μC Microcontroller
-1-
1. Overview
1.1. Introduction
The sun provides a huge amount of energy in the form of radiation. This energy is so
high that it would need 150 Mio. nuclear power plants to reach the same amount. More
precisely, the sun radiation supplies an average power over the entire earth of 164 W
per m2 over a 24 hours day [1]. Since environmental issues caused by human energy
needs are becoming a more important role by leading industry nations, PV is now part
of many renewable energy sources besides wind, water, geothermic energy or biomass.
The energy production by photovoltaic worldwide grew from one Gigawatt in the year
2000 to over 400 Gigawatt in 2017 and will approximately rise to 500 GW in 2018
[2]. This growth is justified. PV development is very important for future generations
to save the energy need of millions of people when fossil energy sources are getting
rarer. Therefore, many universities support projects, where prospective engineers can
work on PV based tasks to deal with future problems and get a basic understanding of
solar energy. This project exposes the use of Maximum Power Point Tracking
Systems.
1.2. Requirements
After the goal of the project was set, the requirement list has been elaborated. These
requirements are divided into a demand, wish or recommendation and are shown in
Table 1. A complete document can be found in the appendix.
The photovoltaic panel is a Size of the PCB STM Board with DC/DC
PV logic STP45 over PINs connected
Implement a MPPT Dimensioning of DC/DC Voltage source for STM
algorithm MPPT as a SEPIC DC/DC Board is the battery
Max. Power consumption of
the PV panel
PCB for power converter
Dimensioning of PCB for
max current of PV panel
DC/DC controllable over
MCU
DC/DC Output constant for
charging Battery
4 Implementation milestone
Program and simulate the concept DJ 22.10.18 13 04.11.18 not done
Program MCU DJ 29.10.18 13 11.11.18 not done
Design PCB layout DJ 29.10.18 20 18.11.18 not done Deadline for ENIB's order: 16.11.18!!!
Verify and Print PCB DJ 05.11.18 13 18.11.18 not done
5 Testing/Verifying milestone
Test programmed MCU DJ 19.11.18 6 25.11.18 not done
Test printed PCB DJ 19.11.18 6 25.11.18 not done
Test PV Panel DJ 19.11.18 6 25.11.18 not done
Test battery DJ 21.11.18 4 25.11.18 not done
Test complete Setup DJ 21.11.18 9 30.11.18 not done
Complete List of needed parts DJ 19.11.18 13 02.12.18 not done
6 Optimize milestone
Sort errors by priority DJ 26.11.18 6 02.12.18 not done
Optimize Setup by correcting the errors DJ 26.11.18 6 02.12.18 not done
2. Theoretical Principles
In this chapter, the principles of solar energy, photovoltaic, Maximum Power Point
Tracking and DC/DC converter are explained. In the end, it is summed up by giving
an example of how to control the output of a photovoltaic panel.
through the atmosphere: the lower the sun stands, the longer the way through the
atmosphere, the more the influence of weak-processes [3].
Figure 2: Mean monthly sums of daily direct and diffuse radiation in Kassel,
Germany [3]
2.2. Photovoltaic
Photovoltaic describes the process of generating electrical power from insolation. This
is done by photovoltaic cells which are parallel- or series connected to create a module
(see chapter 2.2.1). Today, this technique is not only used in small products like
calculators, lamps or toys but also for the energy needs of houses on rooftops or even
in bigger power plants like the solar star in Rosamond, California.
A solar cell converts the energy of light into electricity. When the photons in sunlight
hit the solar panel, positive and negative charge carriers are released in the cell
(compare with Figure 3). Via a semiconductor the light or heat can be forwarded. An
ideal semiconductor is silicon, which occurs in quartz sand, is inexpensive and can be
made highly pure and monocrystalline. The semiconductor is incorporated with
foreign atoms, thereby the positive and negative charge carriers are forwarded
correctly. The p-doped silicon layer with trivalent boron atoms has a positive charge
surplus, the n-doped layer with pentavalent phosphorus atoms is negative charged.
-4-
Between the p- and the n-doped layer is the boundary layer or p-n junction. Here the
impact of the two layers creates an internal electric field. When the photons of sunlight
penetrate, electrons are separated from their atoms. These migrate into the n-layer. The
vacated “electron holes” move towards the p-layer. At the front and back of the cell,
the electrons are dissipated via metal contact layers (mostly aluminum or silver) and
then they flow as current through an electrical conductor [3] [4].
A photovoltaic module has several solar cells connected in series to increase the output
voltage of the module. Commercial crystalline solar modules have between 36 and 144
solar cells which lead to a performance between 170 and 300 W.
The front of module is made of glass (see Figure 4). It must withstand hailstorms and
should also have the highest possible transmittance. That is why low-iron white glass
is normally used. Therewith transmittance with about 95 percent are reached. The solar
cells are embedded between two EVA (Ethylenvinylacetat) foils for mechanical
stabilization, electrical insulation and protection against the effects of the weather. The
synthetic foil of the back of the module consists of PVF (Polyvinylfluorid) and PET
(Polyethylenterephthalat), which ensures the electrical insulation and the UV
resistance [3].
• Standalone systems
• Grid-connected systems
Stand-alone photovoltaic power systems are independent of the utility grid and may
use solar panels only or may be used in conjunction with a diesel generator, a wind
turbine or batteries.
Solar hybrid power systems combines solar power from a photovoltaic system with
another power generating energy source. A common type is a photovoltaic diesel
hybrid system, combining photovoltaics (PV) and diesel generators, or diesel gensets,
as PV has hardly any marginal cost and is treated with priority on the grid. The diesel
gensets are used to constantly fill in the gap between the present load and the actual
generated power by the PV system [5].
An ideal PV cell can be described as a current source with a parallel diode (see
Figure 5). However, metal contacts (Rs) and leakage of the PN junction (Rsh) must
be considered in the practical circuit.
For calculating the output current, the photo-current and saturation current is
necessary.
Module photo-current:
The module saturation current I0 varies with the cell temperature, which is given by
(Equation 4) 𝑇 3 𝑞 ∗ 𝐸𝑔0 1 1
𝐼0 = 𝐼𝑟𝑠 [ ] exp[ { − }]
𝑇𝑟 𝐵𝑘 𝑇𝑟 𝑇
I0 = saturation current
A = B = ideality factor
The top diagram shows that the short-circuit current ISC is almost proportional to the
sun radiation. Besides that, the open-circuit voltage UOC is more affected by the
temperature of the cell then a changing insulation (see bottom diagram). The electrical
power, which is calculated by the product of current and voltage, has a definitive
maximum which is called Maximum Power Point (MPP). In Diagram 1, this MPP is
marked by short circles. From this it follows that for each level of insulation and
temperature, the photovoltaic cell has its optimal operating point, the MPP, at different
locations. Therefore, to get the highest power consumption from the cell, the MPP
must be calculated at any time and proximately, change the behavior of a connected
circuit (see chapter 2.4).
Diagram 3: U-I- (red) and P-U-curve (blue) of a solar panel with U-I-square [6]
- 10 -
Diagram 5: I-V and P-V curve under partial shading condition [7]
2.3.1. General
In Table 2 five commercially available DC/DC converters are compared based on five
decisive features. The three established converters Buck-Boost, SEPIC and Flyback
are described in the following chapters. The circuit of the Cúk converter is the same
as the SEPIC’s, only the inductor L2 and the diode are interchanged.
2.3.2. Buck-Boost
While in the On-state, the input voltage source is directly connected to the inductor L
and energy accumulates in L. In this stage, the diode locks and the capacitor supplies
energy to the output load (see Figure 8).
While in the Off-state, the inductor generates an opposite induction voltage, the diode
becomes conductive, so energy is transferred from L to C and R.
Figure 10 shows an example, where the input voltage is 10 V and the output voltage
is -2,48 V with a duty cycle of 1:4. When MOSFET Q is switched on, the inductor is
charged and the inductor current rises, when Q is off, the inductor is discharged and
its current increases.
- 13 -
2.3.3. SEPIC
Figure 11 shows a simple circuit diagram of a SEPIC converter. The essential elements
of the circuit are three energy storages (the inductors L1a and L1b and the capacitor
CP) and the MOSFET Q1. There are two different switching phases. In phase one Q1
is on (see Figure 12) and in phase two Q1 is off (see Figure 13). Before the MOSFET
Q1 is switched on, the capacitor CP is charged to the input voltage, VIN, the output
voltage Vout is 0 V, no power flows in any of the components.
- 14 -
When Q1 is switched on, the voltage across L1a is VIN. The current across the inductor
increases in form of a ramp, therefore energy is stored in L1a. How steep IL1a is,
depends on the input voltage and the inductance of L1a. As a result, the capacitor CP
charged on VIN, it acts as input voltage for L2, in which a current flows across L2 and
the energy of the capacitor is transferred to the inductor. In this time the diode D1
works in the reverse direction and the capacitor Cout must provide the current for the
connected load.
In the second phase, the MOSFET is off and the polarity at the inductor’s changes.
The diode conducts the saved energy to Cout and the load. Therewith current can still
flow through L1a, the voltage of Q1 must be VIN + Vout (if the voltage of the diode is
not disregarded, it must be still added). The current, which flows through CP charge it
again to VIN.
- 15 -
Figure 14 shows the voltage of Q1 and the inductors. When Q1 is off, the voltage
across L1b must be VOUT. Since CIN is charged to VIN, the voltage across Q1 when Q1
is off is VIN + VOUT, so the voltage across L1a is VOUT. When
Q1 is on, capacitor CP, charged to VIN, is connected in parallel with L1b, so the voltage
across L1b is –VIN.
- 16 -
The currents flowing through various circuit components are shown in Figure 15.
When Q1 is on, energy is being stored in L1a from the input and in L1b from CP. When
Q1 turns off, L1a’s current continues to flow through CP and D1, and into COUT and
the load. Both COUT and CP get recharged so that they can provide the load current and
charge L1b, respectively, when Q1 turns back on [11].
- 17 -
Duty cycle
Assuming 100% efficiency, the duty cycle, D, for a SEPIC converter operating in
CCM is given by
(Equation 5) Vout + VD
D =
Vin + Vout + VD
D = Duty Cycle
2.3.4. Flyback
On-state:
When the switch is closed (see Figure 17), the primary of the transformer is directly
connected to the input voltage source. The primary current and magnetic flux in the
transformer increases, storing energy in the transformer. The voltage induced in the
secondary winding is negative, so the diode blocks. The output capacitor supplies
energy to the output load.
Off-state:
When the switch is opened as in Figure 18, the primary current and magnetic flux
drops. When the magnetic flux is reduced, a voltage of opposite polarity is produced
in the secondary, the diode becomes conductive and current flows from the transformer
to the capacitor and the load.
There are several algorithms with which it is possible to find the MPP. In the following
chapters, some techniques have been selected and categorized. Furthermore, they are
shortly described. In Table 3, a list of all researched MPPT techniques can be found.
- 20 -
Analog/ Implementation
Nr. Tracking method Digital Sensors Speed complexity
1 Constant voltage A V Fast Easy
2 Open-circuit voltage A V Fast Easy
3 Short-circuit voltage A C Fast Easy
4 Open-circuit voltage pilot PV cell method A V Fast Easy - Medium
5 Temperature gradient (TG) algorithm A V& T Fast Easy - Medium
6 Temperature parametric (TP) method A T& I Fast Medium
7 Feedback voltage/current method A V or C Slow Easy
8 P-N junction drop voltage tracking technique A V Fast Medium
9 Look-up table method D I&T Fast Easy
10 Load current/voltage maximization A V or C Slow Easy
11 Linear current control (LLC) D V& C Slow Easy
12 The only current photovoltaic technique A C Slow Easy - Medium
13 PV output senseless (POS) control D C Slow Easy
14 Perturb and observe (P&O) A/D V& C Slow Easy
15 Modified P&O with fixed perturb D V& C Slow Easy - Medium
16 Conventional P&O with adaptive perturb D V& C Fast Easy - Medium
17 Modified P&O with adaptive perturb D V& C Fast Easy - Medium
18 Variable step size P&O D V& C Fast Easy - Medium
19 FLC based P&O D V& C Slow Medium
20 PSO based P&O D V& C Fast Medium
21 Three-point weight comparison D V& C Fast Easy
22 On-line MPP search algorithm D V& C Slow Medium
23 DC-link capacitor droop control D V& C Slow Medium
24 State-based MPPT method D V& C n/a n/a
25 Linear reoriented coordinates D V& C Slow Easy - Medium
26 Curve-fitting technique D V Fast Medium
27 Differentiation method D V& C Slow Medium
28 Slide control method D V& C Very fast Easy - Medium
29 Current sweep D V& C Slow Medium
30 dP/dV or dP/dI feedback control D V& C Slow n/a
31 Incremental conductance (INC) D V& C Slow Easy - Medium
32 Variable step INC D V& C Fast Medium
33 Fractional order INC method (FOICM) D V& C Fast Medium
34 Power increment - INC D V& C Slow Easy - Medium
35 Variable step-size incremental-resistance D V& C Fast Medium
36 Parasitic capacitance D V& C Fast Medium
37 β-method D V& C Very fast Easy
38 IMPP/VMPP computation D V, C, T & I Fast Medium
39 Methods by modulation D V& C Slow Medium - Complex
40 Ripple correlation control (RCC) D V& C Slow Complex
41 Fuzzy logic control (FLC) D V& C Very fast Complex
42 Neural network (NN) / Artificial NN (ANN) D V& C Very fast Complex
43 Biological swarm chasing algorithm D V& C Very fast Complex
44 Extremum Seeking Controller (ESC) D V& C Fast Complex
45 Distributed MPPT (TEODI controller) A I n/a Prototype
46 Bisection search theorem (BST) MPPT D V Slow Easy
47 System oscillation control D V Fast Complex
48 Array reconfiguration A n/a Slow Complex
49 Partial Shading Methods D V, C, T & I n/a Very complex
These methods are using predefined fixed values that characterize the MPP. A few are
proximately described.
This method is based on the linearity between the PV current at MPP and the short-
circuit current. In fact, these currents are proportional to each other by a factor kSC.
This k-value depends on the characteristics of the PV array. The equation is shown
below.
First, ISC is measured periodically with a current sensor by switching of the PV array
for a short time. Then IMPP is computed, the PV panel is reconnected, and the algorithm
starts again after a specified waiting time (see Figure 19). To prevent power loss of the
shutdown event, a modified version of this algorithm calculates ICS for different values
- 22 -
This method uses a pilot cell to measure the k-factor which was described in the two
previous shown algorithms. A feedback loop changes the MPP by using the values
kOC/kSC. It improves the system by avoiding a complete shutdown of the PV generator
to measure the factors but unfortunately, this method gets inaccurate if the PV array
gets bigger (partial shading/mismatch between pilot cell and array) [5].
The open-circuit voltage depends linearly on the cell temperature (see Equation 8).
Thus, the temperature of the PV cell can be measured, VOC calculated, and VMPP
computed with the aid of the k-factor (see above) [5].
Equation 8 𝑑𝑉𝑂𝐶
𝑉𝑂𝐶 ≅ 𝑉𝑂𝐶𝑆𝑇𝐶 + (𝑇 − 𝑇𝑆𝑇𝐶 ) ∙ ( )
𝑑𝑇
- 23 -
As the name indicates, these methods are using attempts in changing parameters of the
system. The following result determines the direction criteria for the next attempt to
reach the MPP. A few are proximately described.
This method is only usable for large PV generation systems. The power conversion is
controlled by a PWM signal which is sent to a DC/DC converter and the MPP is
calculated just by the observation of Iload. It assumes that the maximum current
consumption is proportional to the maximum power consumption. The control circuit
is easy to realize and can be seen in Figure 22. It compares the old with the new duty
cycle values and the old with new load current values and changes the duty cycle
depending on the result [5].
This method runs periodically by perturbing the PV terminal voltage and comparing
the PV output power in three points of the PV curve. Depending on this comparison,
the status is set to positive or negative weighting which leads to a decision whether the
MPP is left, right or at the same position than the perturb point. The process runs as
follows: PA is calculated at duty cycle A followed by setting duty cycle B to duty cycle
A + 1. PB is calculated at duty cycle B whereupon duty cycle C is set with duty cycle
A – 1. Then, PC is calculated, and all results are compared (see Figure 23) [5] [14].
Figure 23: Possible cases of power variation for the three-point weight comparison
algorithm [5]
First, PPV is calculated by measuring VPV and IPV. Then, Perror = Pactual - Pref. If now
Ptolerance is greater than Perror, “actualref” values of P, V and I are changed (see Figure
24) [5].
- 26 -
This method is specified for the use with an AC inverter. The input voltage of the PV
panel VPV and the output voltage Vlink of the DC/DC boost converter are related by
Equation 10. When the maximum PV power is exceeded, Vlink across the capacitor
starts to droop. Right before this point Ipeak is at maximum level and MPP is found. A
typical block diagram can be seen in Figure 25 [5] [14].
Equation 10 𝑉𝑃𝑉
𝐷 = 1−( )
𝑉𝑙𝑖𝑛𝑘
The available data of the PV system is used by mathematical equations to define the
location of the MPP. A few methods are proximately described.
This method requires the exact knowledge of the physical parameters and
specifications of the PV panel for all climatological conditions. The P-V characteristics
curve can be described by Equation 11. Its coefficients must be predetermined by
measuring samples of PV power, current and voltage in a given interval. Therefore,
the maximum power voltage can be calculated by Equation 12. This method needs a
lot memory capacity and enough processor speed fir the mathematical equations [5].
In this method, the MPP is found by using the determination of the derivative of the
panel output power, where the panel current is manipulated as a decaying exponential
sweep function [IPV(t) = f(t)]. In this function of time (see Equation 14, where c is a
constant of the general solution), the I-V characteristics is obtained and updated at
fixed time intervals. The VMPP can then be computed from the curve. For a detailed
description of the method see [5].
Equation 14 𝑡
𝑓(𝑡) = 𝑐 ∙ 𝑒 𝑘
- 28 -
(17) β-method
The algorithm works as follows: First, VPV and IPV are measured and β is calculated.
When β is in range, the MPPT switches to another MPPT method (e.g. Perturb and
Observe), otherwise, it changes the duty cycle and repeats the process (see Figure 26)
[5] [15].
Equation 15 𝐼𝑃𝑉
𝛽 = ln ( ) − 𝑐 ∙ 𝑉𝑃𝑉 = ln(𝐼𝑆 ∙ 𝑐)
𝑉𝑃𝑉
Equation 16 𝑞
𝑑𝑖𝑜𝑑𝑒𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡: 𝑐 =
(𝐴 ∙ 𝑘 ∙ 𝑇 ∙ 𝑁𝑆 )
This method calculates VMPP and IMPP based on differentiation of PPV including VPV,
IPV, the temperature [T], the series resistance of the module [RS], the temperature
coefficient of the open circuit voltage of the module [β], the number of series and
- 29 -
parallel connected modules [NS, NP] and the curve correction factor [k] (see Equation
17 and Equation 18) [5].
Equation 17 𝑁𝑃 𝑉𝑆 + 𝛽 ∙ (𝑇 − 𝑇𝑟𝑒𝑓 ) + 𝑅𝑠 ∙ 𝐼𝑆
𝐼𝑀𝑃𝑃 = ∙
2 𝑅𝑠 + 𝑘 ∙ (𝑇 − 𝑇𝑟𝑒𝑓 )
Equation 18 𝑁𝑆
𝑉𝑀𝑃𝑃 = ∙ 𝑉 + 𝛽 ∙ (𝑇 − 𝑇𝑟𝑒𝑓 ) + 𝑅𝑠 ∙ 𝐼𝑆
2 𝑆
2.4.5. Measurement and comparison methods
This method uses pre-saved data of temperature and irradiance to compare them with
actual measured values. Based on this comparison, a new VMPP is calculated every
cycle of the system controller timer. Usually, this database needs a large capacity of
memory and is generated on manufacturer specifications or experimental tests [5].
This method assumes that the converter is lossless and either a voltage- or a current
source-type load is connected to the output. It says that maximizing the output power
of the converter maximizes the PV power. For a voltage-type load, the voltage is
constant, and the method tries to find Ioutmax. If the load is a current-type, the current is
constant, and the method searches for Voutmax. One sensor at the output measures V or
I and a feedback control loop changes the PV output conversion to achieve a maximum
[5].
The linear current control method uses the linear relationship between IMPP and the
irradiance level to find the MPP (see Figure 27). IMPP is found by sensing the irradiance
and a PI controller changes Ipnl to this value. For a detailed description of the used
equations, see [5] [14].
- 30 -
Figure 27: Linear approximation between IMPP and the irradiance level [5]
To find the MPP with the neural network technique doesn’t require a detailed
knowledge about the PV parameters. It instead uses parameter approximation. A
neural network consists of three layers: input, hidden and output layer. All these three
layers are connected by nodes (see Figure 28). Input variables can be PV array
parameters like VOC, ISC, temperature or irradiance, whereby the hidden layer consists
of the algorithm to find the MPP. It is used to propagate the input signals to the output
layer based on the transfer function applied on it. The output variable is usually the
duty cycle signal [5] [16].
A partial shading condition can cause low power output from a PV system.
Conventional MPPT techniques sometimes fail to track the MPP due to changed curve
characteristics with many local maxima (see chapter 2.2.4). Optimized maximum
power point-based techniques try to avoid this influence and are able to find the MPP
under partial shading.
Although the goal of this work is to implement a MPPT algorithm to charge a battery
with only one single PV module, wherefore partial shading won’t be a problem, a few
MPP techniques are explained in the following paragraphs.
This method improves the problem of PSO (see above) to find the right duty cycle in
a rapidly changing irradiance condition. It uses a formula (see Equation 19), which
calculates the duty cycle using a constant factor K1. This solves the problem in which
the particle search area gets bigger when the duty cycle is large. The parameter K1
includes results of a reduced search area which improves the efficiency of the system
and guarantees the detection of the global MPP [16].
Equation 19 1
𝑑𝑛𝑒𝑤 = 𝑑𝑜𝑙𝑑 − ∙ (𝑃𝑜𝑙𝑑𝑀𝑃𝑃 − 𝑃𝑀𝑃𝑃 )
𝐾1
(26) Deterministic particle swarm optimization
The following method is specified for the use under partial shading condition. It
improves the conventional PSO by using a modified velocity equation (see Equation
20) and inertia weight (w) as a parameter to generate the duty cycle. A detailed
description of the formula is given in [16].
- 32 -
The Firefly Algotihm (FA) sees the different MPPs of the modules of a PV systems as
fireflies which follow three fundamental rules:
• All fireflies are unisex and will move toward the brighter and more attractive
ones until all of them have been compared (except for itself)
The shown flowchart (see Figure 29) gives an example of how a FA could be
implemented. The goal is to find Vref, which is usually equal to the panel voltage in a
PV system. The values are defined as follows [17]:
This method uses three different groups, described as bees, which have special tasks
to find the MPP under partial shading. Thereby, the optimal power is described as the
food source and the food position is the searched duty cycle. First, there is the
employed bee. It is used to identify the particular food source and share the information
with the others. It also acts as a scout bee to check the new food location. Second, the
unemployed bee is used to gather information from employed bees and spot the
optimum food location. Last, there is the scout bee. It is used to carry the random
search for target the new food source. Employed and unemployed bee groups use an
algorithm to track the target and find the global MPP for the PV system [7].
- 34 -
In this project, five algorithms have been selected to be explained in detail: Constant
Voltage Method, Open-Circuit Voltage Method, Perturb & Observe Algorithms,
Incremental Conductance Methods and Fuzzy Logic. These techniques are commonly
used in PV systems and are described in most of the given literature about Maximum
Power Point Tracking.
The constant voltage method is one of the simplest MPPT techniques. It uses one
voltage sensor to measure VPV to set the duty cycle of the DC/DC converter. A fixed
reference voltage equal to the VMPP must be pre-defined and is dependent on the PV
panel characteristics. First, the algorithm (see Figure 30) measures VPV and compares
it to the reference voltage. If the error is bigger to 0, the voltage will be reduced by a
constant C, otherwise increased. Due to different geographical condition, this constant
needs to be changed [5].
The open-circuit voltage method is like the short-circuit current method. It is based on
the linearity between the PV voltage at MPP and the open-circuit current. This relation
- 35 -
This shutting down process can lead to a power loss, which is possible to avoid by
using a modified method. Here, VOC is calculated for different values of insolation and
temperature. It is required to know the current climate and panel condition
(irradiance/temperature) to set the open circuit voltage. Besides that, the converter
doesn’t need to be shut down [5] [13].
Perturb and Observe (P&O) based MPPT algorithms are one of the most used methods
described in literature. It is a general technique to find the duty cycle for a power
converter at the maximum power point of the PV panel by a trial and error process.
The basic P&O algorithm works as follows (see Figure 32). First, VPV and IPV are
measured and a microcontroller calculates the current PPV. Proceeding from a previous
measurement cycle, the power delta ∆P = PPV - PPV_old and voltage delta ∆V = VPV -
VPV_old are calculated. If the power increases, the perturb direction is kept in the same
direction. Otherwise, this operating voltage is perturbed in the opposite way. The MPP
is reached, when ∆P/∆V = 0. The perturbation process is either done by a direct change
of the duty cycle or a change in voltage. If the algorithm output is a change in voltage,
a PID controller modifies the duty cycle to achieve the voltage variation.
- 36 -
On the one hand, this method requires no knowledge of the PV characteristics and it
can be implemented for every kind of PV system. On the other hand, it is very
dependent on its perturbation parameters. A large step size decreases the process time
to find the MPP, but unfortunately increases the oscillation around that point. The
method parameters must be set individually for the specific system, which is used to
achieve best results. To improve this problem, in modified P&O algorithm for
example, the step size is changed depending on the power change in a cycle [5] [13]
[18] [14] [18] [7].
Besides P&O algorithms, incremental conductance (INCC) based MPPT methods are
a widely used technique for PV power systems. This method is based on the fact that
the slope of the PV array power versus voltage (P-V-curve) is zero at the MPP. Like
the P&O technique, INCC first measures VPV and IPV to get the actual power input to
the system, but as opposed to P&O it uses mathematical equations to know, if the
current duty cycle is too low, too high or right at MPP (see Equation 24, Equation 25
and Equation 26). For example, if the change in current over the change in voltage is
higher than the negative current over voltage, the duty cycle needs to be decreased,
which leads to a higher panel voltage. A typical flowchart can be seen in Figure 33 [5]
[16].
- 37 -
Equation 24 𝑑𝑃 𝑑𝐼 𝐼
= 0𝑤ℎ𝑒𝑛 = − (𝑎𝑡𝑀𝑃𝑃)
𝑑𝑉 𝑑𝑉 𝑉
Equation 25 𝑑𝑃 𝑑𝐼 𝐼
> 0𝑤ℎ𝑒𝑛 > − (𝑙𝑒𝑓𝑡𝑜𝑓𝑀𝑃𝑃)
𝑑𝑉 𝑑𝑉 𝑉
Equation 26 𝑑𝑃 𝑑𝐼 𝐼
< 0𝑤ℎ𝑒𝑛 < − (𝑟𝑖𝑔ℎ𝑡𝑜𝑓𝑀𝑃𝑃)
𝑑𝑉 𝑑𝑉 𝑉
To avoid the problem mentioned above, which occurs due to fixed step size values
(see chapter 2.5.3), the INCC method can be modified by a variable step size tracker,
which calculates the step size value automatically depending on the distance between
the operating point and the MPP: the bigger the slope, the bigger the step size (see
Equation 27). Besides that, a small marginal error can be added to the MPP condition
(see Equation 28). This makes the algorithm more stable. If the operating point is not
at the technical correct MPP but very close, it will stay and wait for a bigger change
[5] [16].
Equation 28 𝑑𝐼 𝐼
|( ) + ( )| ≤ 𝜀
𝑑𝑉 𝑉
Different types of this algorithm can be found in [16]. A few examples are:
Besides commonly used trial & error- and mathematical based MPPT techniques, an
often implemented intelligent MPPT method is called fuzzy logic. A Fuzzy Logic
Controller usually consist of three stages: fuzzification, Interference method (rule base
lookup table / fuzzy knowledge base) and defuzzification. Inputs of such a system are
an error E and a change in error ∆E (see Equation 29 and Equation 30). The output is
the change in duty cycle (∆D) for the converter of the PV power system. The result of
E(n) gives the controller information if the operating point is left or right of the MPP
and ∆E expresses the displacement direction. The three stages are explained below.
The fuzzification converts numerical input variables into linguistic variables such as
PB (positive big), PS (positive small), ZE (zero), NS (negative small), NB (negative
big) based on membership functions (see Figure 35). The interference method contains
a rule table with which the fuzzy controller chooses the output based on IF THEN
statements (see Figure 34). For example, IF E(n) is PB AND ∆E is ZE THEN ∆D is
PB → if operating point is distant from MPP towards left hand side and the change of
slope in P-I-curve is about zero, increase duty ratio largely. Last, the defuzzification
transforms fuzzy information into deterministic information. Here are three methods
possible: center of area (COA), mean of maxima (MOM), max criterion method
(MCM) [5] [13] [14] [18].
- 39 -
Rechargeable batteries always consist of two electrodes, cathode and anode, immersed
in an electrolyte solution, as shown in Figure 36. The different battery types differ by
the electrode material used and the electrolyte. At a lead-acid battery both electrodes
are composed of lead or a lead alloy. The electrolyte used is sulfuric acid. The
differently polarized electrodes are separated by a separator and prevent a short circuit.
In the fully charged state, the negative electrode (anode) consists of lead, the positive
electrode (cathode) consists of lead dioxide. The electrolyte stores most of the
chemical energy. At discharging, sulfuric acid, lead and lead dioxide connect and react
with each other and release lead sulphate, water and electrical energy. At charging this
reaction returns. The inner mechanic structure of solar lead-acid batteries is optimized
for a very high lifetime, cycle stability and the behavior at deep discharging. 1200
cycles are typical until a rest capacity of 80 % [19].
- 40 -
The process to charge a 12V battery from a solar panel can be typical divided into four
main stages:
In this state, the battery charge is below the discharge threshold. It has been
deeply discharged or a shorted cell. A charging should start with a low trickle
current and the voltage must stay below the discharge threshold if the battery
has shorted cells. If all cells are ok, the battery voltage climbs until a change to
bulk charging is possible.
A charger acts as a current source in this stage. The PV tries to charge the
battery by providing a constant charge current and operates at the MPP. When
the battery voltage gets to ~95% of the Over Charge Voltage, the charger
switches to the Over Charging State.
The Over Charging State charges the last missing capacity of the battery in a
minimum of time and avoids over charging at the same time. As soon as the
Over Charge Voltage is reach, the charge current decreases immediately.
- 41 -
After the battery got charged to its highest capacity, the Float Charging State
keeps the battery voltage at float voltage level. The charger uses as much
current as needed, to bring the battery voltage to Vfloat. The goal is, to protect
the battery against self-discharge. In the moment, when the battery voltage
drops below 90% of the float voltage, the charging algorithm switches to bulk
state.
These four stages are just an example on how to charge a battery by a PV system
correctly. There are different ways or combinations of the mentions techniques to
achieve similar results.
2.7.1. Overview
A typical MPPT setup consist of four main blocks (see Figure 37): a PV module, the
power converter, an MPPT with PWM generator and the load. The size of the PV
module can be either one little cell, one PV panel or even an PV array (several in series
connected PV panels). Further, the converter depends on the required output of the
system and is either DC/DC or DC/AC. Different converters are shown in Table 2.
Next, the MPPT contains a PWM signal which is sent to the transistor of the converter
to control it. Usually, a microcontroller or digital signal processor is programmed with
an MPPT algorithm. The tracker can also be equipped with voltage-, current-,
temperature- or irradiance sensors [5].
- 42 -
“The Solar Explorer kit provides a flexible and low voltage platform to evaluate the
C2000 microcontroller family (Piccolo F2803x controlCARD) for variety of solar
power applications.” [20]
The board contains a photovoltaic panel emulator and different power converters,
including a DC/DC Boost-, DC/DC SEPIC- and DC/AC Single Phase Inverter (see
Figure 38). For quick demonstration purposes and first understanding of the functions,
the software packages includes, besides a GUI to show/validate the PV emulator work
procedure (see Figure 39), several example codes for the use of MPPT techniques. One
example is “implementing a photovoltaic battery charging system” (see Figure 40).
This sample code will be used, to understand the basic workflow of an MPPT
algorithm by using a SEPIC-converter to charge a battery with a 50 W PV panel (see
chapter 2.7.3).
- 43 -
Figure 38: Texas Instruments Incorporated Solar Explorer Kit (top view)
Figure 41 illustrates the control of the battery charging system, along with the LED
control using the DC/DC SEPIC stage and the DC/DC boost stage. In such systems,
during daylight the battery gets charged and in the evening the charge from the battery
is used to drive the LED’s.
Figure 42: DC/DC SEPIC for battery charging with MPPT [20]
Figure 42 shows the DC/DC SEPIC for battery charging with MPPT and Figure 43
illustrates the control proposed for this stage when doing MPPT. The control when
doing MPPT is like the boost stage however when the battery is not in the bulk
charging stage the MPP cannot be maintained as the battery cannot absorb the max
power from the panel. Hence the control of the stage changes from input voltage of
the stage/ output of the panel regulation to the output voltage of the stage regulation.
The instance when the control is switched is dependent on the battery type and
charging algorithm [20].
In Figure 44 and Figure 45 two MPPT algorithm code examples are shown: The
P&O- (first Figure) and the INCC method (second Figure). Both codes are content of
the Solar Explorer Kit and are written as a macro define. The output voltage is
forwarded to a 2pole2zero1 feedback control loop, to change the duty cycle
1
Similar to PID controllers: The 2P2Z is best suited to phase compensation, where pole & zero
frequencies must be adjusted to meet specifications on the open loop frequency response [21].
- 46 -
depending on the calculation of the algorithm. The examples show that the P&O
method is using less calculations than the INCC technique.
3. Concept
In this chapter, the developed results of this project are presented. First, based on a
research on different MPPT algorithms, some techniques are chosen to be part of the
concept. Then, the battery storage, DC/DC converter and STM Nucleo board are
selected and presented. In the end, the developed power converter are shown and its
featured explained.
- 48 -
All showed MPPT techniques have been already classified and divided into different
chapters based on their algorithms. Code examples in how to implement the P&O and
INCC methods are given by Texas Instruments in the explained solar explorer
development board (see chapter 2.7.3). Also, these two algorithms are one of the most
used techniques for PV systems described in literature. Therefore, these algorithms
shall be developed and compared for the DC/DC converter concept in this project.
Besides that, a simple third method shall be developed and implemented, based on the
knowledge of the research work. This simple method shall work as best as possible
with the DC/DC converter and find the MPP in any weather condition.
For this project, the Yuasa REC10-12 (12 V, 10 Ah) battery can be used. Alongside
costs, this battery is compact, has high charging cycles and a long life. It’s Open-
Circuit Voltage is 15V. The Floating Voltage stands at 13.65 V and it has a Discharge
Threshold of 10.5V. These specifications are close to the recommend values for a
battery with which the TI Solar Explorer Kit can be used.
The Nucleo-F303RE provides four ADCs, which can independently be used (see
Figure 49). The board can easily be programmed by using the STM32CubeMX
software to define input/output ports, timer declarations and user labels. It
automatically generates the program code including all definitions. The chosen HAL
library is provided by STMicroelectronics and is programmed using the free software
IDE “System Workbench for STM32”. A PIN overview is shown in Figure 48. The
STM32 Nucleo board includes all possibilities to design an MPPT system. It is part of
the requirements list.
The layout of the board is close to the SEPIC stage seen on the Solar Explorer Kit (see
chapter 2.7.2). Several tests (see chapter 4) were made to verify the component design
of Texas Instruments. On top, a quick example shall show that all components have
been overdesigned to ensure a good work function. The application report “AN-1484
Designing A SEPIC Converter” by TI is an eleven pages long paper to point out, how
to design and calculate a SEPIC converter see [23]. The following example shows the
selection of the inductors. First, the duty cycle D has to be considered. For a SEPIC
converter in continuous conduction mode, D is given by Equation 31, where VD is the
forward voltage drop of the diode. Thus, Equation 32 calculates the maximum duty
cycle.
Equation 31 𝑉𝑜𝑢𝑡 + 𝑉𝐷
𝐷 =
𝑉𝑖𝑛 + 𝑉𝑜𝑢𝑡 + 𝑉𝐷
- 54 -
Equation 32 𝑉𝑜𝑢𝑡 + 𝑉𝐷
𝐷𝑚𝑎𝑥 =
𝑉𝑖𝑛(min) + 𝑉𝑜𝑢𝑡 + 𝑉𝐷
Equation 33 𝑉𝑜𝑢𝑡
∆𝐼𝐿 = 𝐼𝑖𝑛 ∙ 40% = 𝐼𝑜𝑢𝑡 ∙ ∙ 40%
𝑉𝑖𝑛(min)
AN-1484 describes a rule for determining the inductance. It says to allow the peak-to-
peak ripple current to be around 40% of the maximum input current at the minimum
input voltage (see Equation 34). With a max Iin of 2.7A (according to the datasheet of
the PV panel for short circuit), this equals to a delta IL of 1.08 A.
Equation 34 𝑉𝑖𝑛(min)
𝐿1 = 𝐿2 = 𝐿 = ∙𝐷
∆𝐼𝐿 ∙ 𝑓𝑠𝑤 𝑚𝑎𝑥
Then, with a switching frequency of 100kHz, the inductor value is calculated to 40
μH.
The SEPIC stage of the Solar Explorer Kit contains two 100 μH inductors for close
calculation values which were used above. This is far more than the result shows. Due
to that, the self-designed SEPIC converter of this project uses similar or equal
components as the TI SEPIC converter.
Although the solar kit conduces as a model, there are many differences. First of all,
there is the pin connection to the STM μC. It simplifies the use of the Nucleo board to
control the MPPT mechanisms and reduces cable management. Nevertheless, it has to
be said that this design is tailored to the F303RE. Secondly, a battery status LED bar,
achieved by an IC called LM3914, was added to the board. It gains to see, if the battery
has enough power or if the charging process works. It lights up between 10 – 14 V.
Finally, the battery delivers the energy for the MOSFET driver, the status LEDs and
the Nucleo board. The power supply for the STM is achieved by a 12V to 5V DC/DC
converter by TRACO Power.
The raw and finished board can be seen in and shows the full-fledged converter with
all soldered components.
- 55 -
4. Tests
4.1. Simulations
4.1.1. Simulation of the PV-panel
It was built a PV model in Simulink (see Figure 53). Depending on the irradiation and
the temperature, the PV model puts out the corresponding current-voltage and power-
voltage curve (compare Figure 54). (The theory about the characteristics of solar
panels and its output current and voltage is explained in chapter 2.2.4). The input
parameters of the PV panel irradiation G and temperature T can be changed in the
MATLAB code (see Figure 56). The output value Ipv is calculated in the subsystem
PV panel. The calculation is described in chapter 2.2.3. In (Equation 1 is evident that
the PV current Ipv depends on the PV voltage Vpv. The PV voltage in turn correlates of
the resistance R. The solar panel acts as a current source. With a parallel resistance,
the current source is changed to a voltage source. In this case it is chosen a very high
resistance R of 50 ohm.
Figure 54: current-voltage and power-voltage curve of the Simulink PV-model with a
50 Ohm parallel load resistance
Figure 55: PV current, voltage and power with a 50 Ohm parallel resistance
- 59 -
First it is simulated the SEPIC Converter with a constant voltage source. In the reality
the PV panel is a current source and its current Ipv is not constant. But to test and design
the SEPIC, it is constructing a model with a constant source. The SEPIC circuit is built
in LTspice (see Figure 57) and in Simulink (Figure 58) with the same components as
the TI-SEPIC. Therefore, the measurements of the real TI-SEPIC can be compared
with the LTspice simulations and it can be checked for accuracy. In chapter 5.2.1 the
simulation results are shown: the output voltage and the input current at a variable
Duty cycle of the LTspice and Simulink SEPIC model and the TI-SEPIC.
- 60 -
After the PV model and the SEPIC Converter were simulated, that both Simulink
models are connected and simulated (compare Figure 59).
- 61 -
But in this model is still a mistake. The voltage and current through the 20 Ohm load
resistance at the output of the SEPIC are negative (see Figure 61). Comparing with the
outputs of the SEPIC model with a constant DC source in Figure 60, the results looks
completely different.
Figure 60: outputs of the SEPIC model Figure 61: outputs of the SEPIC model
with a constant DC source with a PV source at a Duty Cycle of
40%
- 62 -
Also the inductor, capacitor and diode currents of the SEPIC model with a PV source
are different as the currents of the SEPIC model with a constant DC source (compare
Figure 62 and Figure 63).
Figure 62: currents of the SEPIC Figure 63: currents of the SEPIC
model with a constant DC source model with the PV source at a Duty
Cycle of 40%
Figure 65 shows that the voltage of inductor L2 of the SEPIC model with the PV source
is always negative and the voltage of inductor L1 is always positive. The inductor
voltages of the SEPIC model with a constant DC source are both negative and positive,
depending if the MOSFET is off or on.
- 63 -
Figure 64: voltages of the SEPIC model Figure 65: voltages of the SEPIC model
with a constant DC source with a PV source
Therefore, the current and voltage of the PV looks correct (see Figure 66).
After the mistake in the first Simulink model with the simscape SEPIC and the PV
source was not found, a second model without simscape components was build (see
Figure 67).
- 64 -
𝑃𝑏𝑎𝑡 = 𝜂 ⋅ 𝑃𝑃𝑉
Equation 35
𝑉𝑃𝑉 𝐼𝑏𝑎𝑡 𝐷
Equation 37 = =
𝑉𝑏𝑎𝑡 𝐼𝑃𝑉 ⋅ 𝜂 1 − 𝐷
Equation 38 𝐷
𝐼𝑏𝑎𝑡 = ⋅𝐼 ⋅𝜂
1 − 𝐷 𝑃𝑉
𝐷
Equation 39 𝑉𝑃𝑉 = ⋅𝑉
1 − 𝐷 𝑏𝑎𝑡
- 65 -
𝜂= efficiency
D = Duty Cycle
Vbat is calculated with a SOC (state of charge) – OCV (open circuit voltage) Lookup
Table (compare Figure 69 and Figure 70).
- 66 -
Figure 72 shows that at a rising Duty Cycle the PV current increase, and the PV voltage
decrease. The battery voltage is always about 12 V and the battery current rise steeply
until a Duty Cycle of 45% and afterwards it decreases. Only from a Duty Cycle of 35%
currents flow. Probably it caused on the efficiency of 90%. When the Duty Cycle is
too small a battery voltage of about 12 V cannot reached. The maximum power is
reached at 45% Duty Cycle.
This test shall show that the PWM control of the Nucleo board is working correctly
and if the SEPIC converter by TI is designed properly. Therefore, its results (see
chapter 5.2.1) shall be compared to an LTspice simulation which is using the same
component parameters. A block diagram of this test is shown in Figure 73.
- 68 -
The flow chart of the program is seen in Figure 74, Figure 75 and Figure 76. The duty
cycle is controlled by a variable resistance, which is connected to an ADC input of the
STM and a 3.3V power source. Depending on the input, the duty cycle can be regulated
manually from 0 to 100%.
Figure 74: Flow chart of main.c for testing the TI SEPIC converter
- 70 -
Figure 75: Flow chart of function user_pwm_calcduty for testing the TI SEPIC
converter
Figure 76: Flow chart of function user_pwm_setvalue for testing the TI SEPIC
converter
- 71 -
The following tests use the TI example code showed in chapter 2.7.3. A battery and an
LED are connected to the Solar Explorer Kit (see Figure 77). First, the PV emulator
of the board is used to verify the function of the example code. All variables can be
monitored on the IDE while the code is running.
In the next test, the PV emulator is turned off and the PV logic panel is connected to
the board instead (see Figure 78). On the hardware, a light sensor detects, if light is
present. This variable can also be changed by using the IDE. During light condition,
the MPPT should work and deliver as much power from the panel to the battery as
possible. If no light is present, the algorithm should turn on the provided boost
converter to light up the connected LED. The required energy is given by the battery.
Measurements are made with an oscilloscope.
- 72 -
This test checks how the STM Nucleo board can be programmed properly to read
correct ADC values and send the required PWM signal to control the SEPIC converter.
Here, the STM is connected to the Dummy controlCard and the PV logic panel plus
battery to the Solar Kit board (see Figure 79 and Figure 80). Different constant duty
cycles shall show that an MPP exist and if the STM can control the converter like the
Piccolo controlCard. As in the previous chapter, measurements are made with an
oscilloscope (see Figure 81).
Figure 79: Block diagram of TI SEPIC testing with PV panel and battery
- 73 -
The first test is made to check general functions of the converter. The STM Nucleo,
PV panel and battery are connected to the board. The test shall proof, if the LED bar
is working correctly, if the power supply for the STM, 12to5DC/DC converter and
MOSFET driver delivered by the battery is working correctly, if all components are
connected right and if the MOSFET driver changes the PWM signal by the STM
accurately.
The second test is done to verify if all functions work properly together. A similar test
is made in chapter 4.2.1, but instead of a constant load, a battery needs to be connected
to the converter. This is due to the power supply of the MOSFET driver, which is
delivered by the battery. A constant voltage DC source of 10V is connected to the
Panel input and the STM is programmed to set a duty cycle from 0 – 100% in 5% when
pressing button B1 on the board.
After all functions have been validated, the last tests shall show that an implemented
MPPT on the STM µC can find the correct MPP and set the calculated duty cycle to
charge the battery with the highest possible power. The general setup can be seen in
Figure 82 and Figure 83.
- 75 -
Figure 82: Block diagram of the test setup to check the developed MPPT system
Two general code structures are used to realize the MPPT system test. The first one is
used to implement the only-current-, P&O and INCC method (see Figure 85 and Figure
86). It uses a 100 Hz interrupt sequence to check the battery state, which is either float-
or bulk-charging (see). If bulk-charging is required, the MPPT method is called by a
macro definition. The advantage of this structure is that, if a different MPPT technique
is claimed, this macro and just a few variable declarations in the code have to be
changed. The output of a method can either be directly the duty cycle or the required
voltage output. If the voltage at MPP is calculated, a PID control loop, realized by an
- 76 -
interrupt routine called with 50 Hz, is changing the duty cycle to reach this value. In
float-charging, the MPPT is off. During this state, the PID controller parameters
changes to the output battery voltage. The output voltage is tried to hold constant on a
defined value.
The second code structure implements a simple MPPT algorithm, which was
developed during the project (see Figure 87 and Figure 88). Every minute, the duty
cycle will go from 0 – 100% in about 20 seconds, calculates at each percent the panel
power and saves the duty cycle value where the highest power was found. A PID
control loop is used for the battery voltage at float-charging and the duty cycle is
directly set by changing the register in the timer.
The tests are made under cloudy and sunny condition. Due to the weather and winter
time, an overhead projector was used as an artificial light source to test the MPP if no
sun was shining (see Figure 84).
- 77 -
Figure 84: Overhead projector to test the MPPT with its DC/DC converter concept
- 78 -
5. Results
This chapter summarizes all results of the tests described in chapter 4. It shows the
outcome of simulation-, TI-SEPIC stage- and self-designed SEPIC converter
inspections. The results are used to understand the given discussion and conclusion in
chapter 6 and 7.
Figure 89 and Figure 90 shows the simulated voltages above the inductors at a Duty
Cycle of 40% and an input voltage of 17.6 V. There you can see, when the PWM
voltage is on, above the inductor L1 is the input voltage and above L2 is minus input
voltage. When the PWM voltage is off, at L1 is the negative output voltage and at L2
the positive output voltage. Therefore, the simulation results accord with the theory in
chapter 2.3.3 (compare Figure 14).
- 82 -
Also the simulated inductor, capacitor, diode and MOSFET current in the LTspice (see
Figure 91) and Simulink model (see Figure 92) correspond with the theory in chapter
2.3.3 (compare Figure 15).
- 83 -
5.1.3. Vout of LTspice and Simulink simulation with a variable Duty Cycle
Figure 93 shows the output voltage of the LTspice and the Simulink SEPIC model at
a variable Duty Cycle. The measured output voltage of the TI-SEPIC is explained at
chapter 5.2.1. Until a Duty Cycle of 70% the output voltage of the LTspice and
Simulink simulation are equal. At a higher Duty Cycle the Simulink output voltage is
higher than the LTspice output voltage. The reason for this behavior is unknown. Both
simulations show that with a rising Duty Cycle the output voltage increase until a
maximum. From a Duty Cycle of 75% (LTspice) respectively 90% (Simulink) the
output voltage decreases. An output voltage between 10 V and 14 V is reached at a
- 84 -
Duty Cycle between 30% and 40%. The voltage is measured over the 20-ohm
resistance.
Figure 93: V_out of LTspice and Simulink SEPIC model and TI-SEPIC with
controlCard Dummy at an input voltage of 17,6 V
The output current of both simulations behaves in the same way as the output voltage
(compare Figure 94). Also, from a Duty Cycle of 70% the curves are different. The
input current changes with a variable Duty Cycle, too. The input current of the LTspice
simulation rises steeply and at a Duty Cycle of 100% it reaches 12,28 A. The
comparison with the input current of the TI-SEPIC is described in chapter 5.2.1.
- 85 -
Figure 94: input current of the LTspice and Simulink SEPIC model and the TI-
SEPIC with controlCard DUMMY at an input voltage of 17,6 V
5.2.1. Results for test with constant voltage source (TI SEPIC)
In Figure 93 the output voltage of the TI-SEPIC with controlCard Dummy is compared
with the output voltage of the LTspice and the Simulink SEPIC model at a variable
Duty cycle. The output voltage of the TI-SEPIC is approximately the same as the
voltage of the LTspice and Simulink simulation. Also, the TI-SEPIC’s input current
acts like the LTspice simulation (compare Figure 94). That ensures that the
measurement of the TI-SEPIC and also the two simulation models work correctly.
Figure 95 and Figure 96 shows that also at an input voltage of 10 V the output voltage
and input current of TI-SEPIC and LTspice SEPIC model are the same. There is no
Simulink simulation for this case, because to simulate the model takes long time.
- 86 -
Figure 95: V_out of LTspice SEPIC model and TI-SEPIC with controlCard Dummy
at an input voltage of 10 V
Figure 96: input current of the LTspice SEPIC model and the TI-SEPIC with
controlCard DUMMY at an input voltage of 10 V
- 87 -
The PV emulator and street light example code of the TI Solar Explorer Kit is used to
understand basic functions of an MPPT system (see chapter 4.2.2). The LED light can
be turned on without any problems by covering the light sensor (see Figure 97). The
PV Emulator can deliver different stages of MPPs. These stages are set by
Gui_LightCommand. For this test, a value of 0.4 is set, which equals an irradiance of
400 W/m2, Pmax of 14.41 W and a maximum power point voltage of 7.363 V. The
battery could be charged until the MPPT switched from bulk- to float-charging (see
“BatteryState” in Figure 98). Therefore, the PV emulator delivers as much power as
required to hold the output voltage at a specific level.
Figure 99 shows the state when no light is detected, and the LED is turned on. It’s
clearly visible that no panel power is delivered by the emulator due to 0% duty cycle
and the battery voltage dropped.
In the next step, the PV logic panel is connected to the TI board and the emulator is
turned off. The P&O and INCC methods of the TI code example are used under cloudy
and sunny conditions.
Under cloudy conditions, both algorithms have problems to find the right MPP on the
first try. After several test repetitions, the panel can deliver a maximum of up to 0.3
W. The best results of the found MPPs can be seen in Figure 100 (P&O: 0.17W) and
Figure 101 (INCC: 0.097W). INCC is faster than P&O but both MPPTs drop to 0%
duty cycle after 10 to 15 minutes.
- 89 -
Figure 100: Battery charging with TI SEPIC, PV panel and P&O algorithm (cloudy)
Figure 101: Battery charging with TI SEPIC, PV panel and INCC algorithm (cloudy)
Under sunny weather conditions, the PV panel delivers a higher power output.
Although the sun stands low in winter, both algorithms have no problems now to find
the MPP after adjusting the “stepsize”-value in the TI code. While testing the P&O
algorithm, the MPP could achieve more than 2.24 W (see). As the sun went down and
some clouds came up, the INCC method could find an MPP with around 0.8 W input
(see). Under this condition, it is possible to measure a visible current flow through the
inductor L2 of the SEPIC converter (see). Its value rises, when the MOSFET switch
(realized by the duty cycle) is “turned on” and decreases as the switch is “turned off”.
From a subjective point of view, the INCC method was a little faster than the P&O
method.
- 90 -
Figure 102: Battery charging with TI SEPIC, PV panel and P&O algorithm (sunny)
Figure 103: Battery charging with TI SEPIC, PV panel and INCC algorithm (mostly
sunny)
- 91 -
Figure 104: Battery charging with TI SEPIC (duty cycle and inductor current
through L2)
In this subchapter, the ADC and code calculations of the STM µC are tested and its
results are presented. The Dummy controlCard is used to control the TI SEPIC over
the STM controller.
The STM is using a 12-bit ADC to measure panel voltage, panel current, battery
voltage and inductor current through L2. An algorithm calculates from these ADC
values the actual voltage or current. The results of the first test under cloudy conditions
with a constant duty cycle of 37.5% are shown in Figure 105.
Figure 105: Testing TI SEPIC over STM with Dummy controlCard (cloudy/dark)
Compared measurements with a multimeter and oscilloscope show that the calculated
values for panel and battery voltage of the ADC are 0.5V too low. A comparison test
with a controlled input voltage to the ADC pins showed that the STM µC is converting
the voltage to the wrong ADC value. Although the difference between V ref- and Vref+
is correct at 3.3 V, the ADC result of the STM with an input of 3.3 V is about 4015
which is 80 steps too low. A detailed description of this result is shown in chapter
- 92 -
5.3.2. Furthermore, the inductor current value is just seen as an approximation. The
real result can only be visible by using an oscilloscope (see Figure 104).
Figure 106: Testing TI SEPIC over STM with Dummy controlCard (sunny)
Same tests are made under sunny conditions to see results with higher power input.
The voltage and current conversions under a constant duty cycle of 32% can be seen
in Figure 106. This measurement leads to the highest power output during the whole
project time. Duty cycle and current flow through inductor L2 on the TI SEPIC
converter shows Figure 107.
Figure 107: TI solar explorer development board test with STM Nucleo control over
Dummy controlCard under sunny conditions
- 93 -
The test of general functions of the converter concept was successful. All connections,
components and power supplies are working correctly, and the LED bar gives the right
feedback of the current battery voltage (see Figure 108).
First tests with the connected PV logic panel under very cloudy conditions show that
the panel voltage drops as the duty cycle increases. Due to the weather, no panel or
inductor current could be measured.
5.3.2. Results for test with constant voltage source (converter concept)
The Panel voltage VPV is calculated of the STM Nucleo with Equation 40.
Equation 40 𝐴𝐷𝐶_𝑣𝑎𝑙𝑢𝑒
𝑉𝑃𝑉 = ∗ 𝑚𝑎𝑥𝑖𝑚𝑢𝑚𝑣𝑎𝑙𝑢𝑒𝑜𝑓𝑣𝑜𝑙𝑡𝑎𝑔𝑒𝑑𝑖𝑣𝑖𝑑𝑒𝑟
𝐴𝐷𝐶_𝑟𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛
The ADC resolution is the maximum value of the ADC (Analog Digital Converter). It
is 4095. The maximum value of the voltage divider is 33.3. It results of the ratio of the
100 kiloohm and 11 kiloohm resistance at the SEPIC input. Applying this value in
Equation 40, it appears for VPV:
𝐴𝐷𝐶_𝑣𝑎𝑙𝑢𝑒
𝑉𝑃𝑉 = ∗ 33.3
4095
- 94 -
But it was determined, that the calculated VPV of the STM Nucleo has always a
deviation to the real VPV value. Probably it causes of the high inner resistance of the
STM Microcontroller. Therefore, the calculation value 33.3 was changed to 35 to
minimize the drift error. The same was made with the maximum value of the voltage
divider to calculate Vbat. The voltage ratio of the two resistances 100 kiloohm and 20
kiloohm results 19.8. But in the STM code this value was matched to 21. To see how
exact the calculated voltage Vbat of the STM is, it was tested with a variable Duty cycle
and compared with the measured value Vbat_multimeter (see Figure 109). The battery was
connected to our SEPIC board and operated with a constant voltage source of 10 V.
At a Duty cycle of 60% it was reached the maximum output current of the DC source.
From this point the source was switched to a constant current source of 1.2 A and at a
rising Duty cycle the source voltage decreased. That explains the high deviation
between Vbat_STM and Vbat_multimeter of 0.6 V at a Duty cycle of 60%. Otherwise the
deviation is within a tolerance of 0.3 V (compare Figure 112).
Figure 110 shows the PV current Iin measured with the STM controller and given by
the DC-source at a variable Duty cycle. At a rising Duty cycle the current increase. At
a Duty cycle of 60% the PV current shoots up. At 1.2 A the current limit of the DC-
source was reached.
- 95 -
Figure 110: Panel current, measured with the STM-microcontroler and given by
DC-source
In Figure 111 you see additionally to the Input current drop, that the input voltage
decreases with a rising Duty cycle and the battery voltage is approximately constant.
Figure 111: our self SEPIC test with DC-source and battery as load and a variable
Duty cycle
- 96 -
Figure 112 shows the absolute deviations of the STM values Vbat, Vin and Iin compared
with the reference values Vbat_multimeter, Vin_DC_source and Iin_DC_source. Iin_STM is mostly
within a tolerance of 0.1 A. Vin_STM is mainly within a tolerance of 1V.
5.3.3. Results for test with PV panel and MPPT (converter concept)
In the following chapter, the results of the tests of different MPPT methods with the
converter concepts are described.
While testing under cloudy condition didn’t generate enough power input (VOCmax =
14V), the overhead projector could achieve a higher open-circuit voltage of up to 18
V. The testing results can be seen in Table 5.
First, the simple MPPT algorithm was implemented and could be tested under sunny
condition and with the artificial light source. This method could always find the MPP
and delivered over 9 W to the battery while the sun was shining. The power
consumption by the overhead projector reached 0.16 W (see Figure 113).
In the next tests, OC, P&O and INCC were programmed to the STM controller. This
time, no sun was shining.
The OC method changed the duty cycle slowly to 100% and stayed at this value. No
MPP could be found.
Further, the P&O algorithm with direct duty cycle change in the macro could find
different duty cycles. The algorithm is dependent on the starting duty cycle of the
program and stayed at 7% when starting between 0 – 20%, stayed at 22% when starting
- 97 -
between 20 – 30%, stayed at 52% when starting between 30 – 50% and found an MPP
at 88% when starting between 60 – 100%.
The INCC method with PID control loop of the calculated voltage change by the macro
always found the same MPP at around 10% duty cycle with 10.2 V and around 0.016
A which is a power input of 0.16 W and equals to the same value found by the simple
MPP at a different duty cycle.
Figure 113: MPP found by own MPPT (overhead projector light source)
- 98 -
6. Discussion
The first goal of this project is to realize a MPPT system for small photovoltaic device.
Therefore, a DC/DC converter was developed and four MPPT algorithms were
implemented as well as tested.
Primarily, the simulation results of the full Simulink model (see Figure 72) shall be
compared to the experimental results of the final converter concept (see chapter 5.3)
and the controlCard Dummy test, where the STM is used to control the TI SEPIC
converter with connected PV panel and battery (see 5.2). When measuring panel
voltage and current as well as battery voltage and inductor current at different duty
cycles with the STM controller, the following results were made under sunny
conditions. At 0% duty cycle, no current is delivered by the panel and its voltage is at
maximum. When increasing the duty cycle, the panel voltage starts decreasing after 2-
3% and, if enough power is delivered by the sun, the panel current start increasing
slowly. As continuing to increase the duty cycle value, there is a point, where the
current start to rise very high while voltage decreases. This percent value is strongly
dependent on sun radiation. After this point, by steadily changing the duty cycle value
by value until 100%, the panel current will languish, and panel voltage will drop to
0V. The Simulink model shows similar results to the above described behavior,
although the panel voltage doesn’t drop at low duty cycle values. The reason could be
power consumption by components of the converter and traces, which are ignored in
the simulation. Another reason could be a mistake in the Simulink model.
Furthermore, the results of a test made with LTspice, MATLAB Simulink and with
the final converter concept are again shown in Figure 114 and Figure 115 to have a
better comparison. Under simulation, a constant voltage input source of 10 V is
connected to the SEPIC converter and a load connected to the battery output (see
4.1.1). The hardware test was made with a DC source connected to the panel input,
but, as the battery supplies the MOSFET driver on the converter concept, instead of
the resistor, the battery was connected (see 4.3.2).
- 99 -
Figure 114: Panel current of converter concept under changing duty cycle with
constant input voltage (10V)
Figure 115: Panel current of converter concept and LTspice simulation under
changing duty cycle with constant input voltage (10V)
The diagrams show that both characteristics of simulation and reality have a panel
current raise at a certain duty cycle and act similar with increasing duty cycle. If no
current limit would be set to protect the components, the curve would look the same.
This means that the DC/DC converter, based on the results of the simulation, is
probably working correctly and provides the same results than the TI SEPIC converter.
- 100 -
One problem during the project came up when using ADCs of the STM Nucleo board.
It converts wrong values wherefore the calculation had to be adjusted (see 5.3.2). The
error could either be because of a false inner resistance or a wrong programming of
the microcontroller. Anyway, the voltage divider resistors provide correct voltage
ratios for panel and battery voltage.
All tests were made under rapidly changing weather conditions in Brest where the sun
didn’t shine frequently. Therefore, a constant condition for comparing different MPPT
algorithms under very low sun radiation is difficult to realize. Even a used overhead
projector could not deliver more than 0.2 W to the converter (see 5.3.3), which is too
low for charging a battery.
7. Conclusion
A DC/DC converter for the given Microcontroller was built. The concept was adopted
of the TI-SEPIC converter. The charging circuit was simulated in Simulink and
compared with tests of the designed converter concepts. The simulations and our
designed converter correspond and show similar results. 49 MPPT algorithm
principles are roughly compared based on 4 features. 28 MPPT algorithm are
proximately described and the five most commonly used MPPT techniques are
explained in detail. Four algorithms are programmed with the Microcontroller. The
battery can be charged by using with the PV panel using our DC/DC converter. At
least one implemented MPPT algorithm is working properly. Therefore, all
requirements for this project are met. Nevertheless, there are improvement suggestions
and many optimization options, like advancing the written code.
All measurements were taken in November and December. During this time, the sun
doesn’t shine very often wherefore solar power wasn’t high enough while testing. The
PV panel is too small (smaller than 0.5 m²) for cloudy conditions. That means that the
MPPT algorithm should be optimized to check, if the currently solar power is enough
for charging the battery and supporting the microcontroller. At night, the battery
should be disconnected of the converter to avoid discharge. All tested MPPTs had
problems to find the right MPP under cloudy conditions or with an artificial light
source (overhead projector). Therefore, the tests should be repeated under sunny
conditions, especially the last implemented INCC method with PID control loop.
Another option is to build a PV emulator. It can emulate the solar power under different
irradiance and temperature conditions and would help to test the MPPT system.
An MPPT system must be chosen by the claimed requirements of the overall system.
A simple algorithm could be enough for a small battery charging system for e.g. a
camping car. More complex intelligent methods, like neural networks or partial
shading techniques, are advisable for bigger panel power plants. The choice of the
microcontroller depends on how precisely an MPPT needs to work. A more complex
algorithm requires calculation power which needs more memory space, clock speed
and a higher energy consumption. Special math libraries make the calculations more
precisely than using float variables. For this project, the STM Nucleo board is the right
choice for a low-cost implementation of MPPT algorithms like P&O and INCC or
similar methods.
The efficiency of solar panels is already fully utilized. Physically, there will be no
great optimization options. And because the efficiency of PV panels is very low, it is
even more important to use the power from the solar module as lossless as possible
and to fully exploit the MPPT potential.
- 102 -
List of Literature
[1] “Solar Energy.” [Online]. Available:
http://zebu.uoregon.edu/disted/ph162/l4.html. [Accessed: 19-Dec-2018].
[2] O. work Rfassbind, English: Exponential growth-curve on a semi-log scale of
worldwide installed photovoltaics in megawatts since 1992. (Figures for
cumulative nameplate capacity in wattDC, watt-peak). 2014.
[3] V. Wesselak and S. Voswinckel, Photovoltaik Wie Sonne zu Strom wird. 2012.
[4] “Wie funktionieren Solarzellen?” [Online]. Available:
https://www.solaranlage.de/technik/solarzellen. [Accessed: 17-Dec-2018].
[5] N. Karami, N. Moubayed, and R. Outbib, “General review and classification of
different MPPT Techniques,” Renew. Sustain. Energy Rev., vol. 68, pp. 1–18,
Feb. 2017.
[6] “White paper which solar charge controller PWM or MPPT,” victronenergy, Jun.
2014.
[7] D. Verma, S. Nema, A. M. Shandilya, and S. K. Dash, “Maximum power point
tracking (MPPT) techniques: Recapitulation in solar photovoltaic systems,”
Renew. Sustain. Energy Rev., vol. 54, pp. 1018–1034, Feb. 2016.
[8] S. J. Chiang, Hsin-Jang Shieh, and Ming-Chieh Chen, “Modeling and Control of
PV Charger System With SEPIC Converter,” IEEE Trans. Ind. Electron., vol.
56, no. 11, pp. 4344–4353, Nov. 2009.
[9] “Prinzip des Sperrwandlers, Inverswandlers und Upconverters.” [Online].
Available: https://elektroniktutor.de/analogtechnik/sperrwdl.html. [Accessed: 11-
Dec-2018].
[10] “Buck–boost converter,” Wikipedia. 31-Oct-2018.
[11] J. Falin, “Designing DC/DC converters based on SEPIC topology,” p. 9,
2008.
[12] “Flyback converter,” Wikipedia. 11-Aug-2018.
[13] P. Bhatnagar and R. K. Nema, “Maximum power point tracking control
techniques: State-of-the-art in photovoltaic applications,” Renew. Sustain.
Energy Rev., vol. 23, pp. 224–241, Jul. 2013.
[14] M. A. Eltawil and Z. Zhao, “MPPT techniques for photovoltaic applications,”
Renew. Sustain. Energy Rev., vol. 25, pp. 793–813, Sep. 2013.
[15] X. Li, H. Wen, L. Jiang, E. G. Lim, Y. Du, and C. Zhao, “Photovoltaic
Modified β-Parameter-based MPPT Method with Fast Tracking,” J. Power
Electron., vol. 16, no. 1, pp. 9–17, Jan. 2016.
[16] S. Saravanan and N. Ramesh Babu, “Maximum power point tracking
algorithms for photovoltaic system – A review,” Renew. Sustain. Energy Rev.,
vol. 57, pp. 192–204, May 2016.
[17] D. F. Teshome, C. H. Lee, Y. W. Lin, and K. L. Lian, “A Modified Firefly
Algorithm for Photovoltaic Maximum Power Point Tracking Control Under
Partial Shading,” IEEE J. Emerg. Sel. Top. Power Electron., vol. 5, no. 2, pp.
661–671, Jun. 2017.
[18] B. Bendib, H. Belmili, and F. Krim, “A survey of the most used MPPT
methods: Conventional and advanced algorithms applied for photovoltaic
systems,” Renew. Sustain. Energy Rev., vol. 45, pp. 637–648, May 2015.
[19] “Rechargeable Batteries - Chemistry LibreTexts.” [Online]. Available:
https://chem.libretexts.org/Textbook_Maps/Analytical_Chemistry/Supplemental
_Modules_(Analytical_Chemistry)/Electrochemistry/Exemplars/Rechargeable_B
atteries. [Accessed: 18-Dec-2018].
- 103 -
List of Figures
FIGURE 1: PROJECT SCHEDULE ........................................................................................................... - 2 -
FIGURE 2: MEAN MONTHLY SUMS OF DAILY DIRECT AND DIFFUSE RADIATION IN KASSEL, GERMANY
[3] .............................................................................................................................................. - 3 -
FIGURE 3: SCHEMATIC OF A PHOTOVOLTAIC SOLAR CELL [3] ........................................................... - 4 -
FIGURE 4: SCHEMATIC OF A PHOTOVOLTAIC MODULE [3]................................................................ - 5 -
FIGURE 5: THE ELECTRICAL EQUIVALENT CIRCUIT OF A PV CELL ....................................................... - 6 -
FIGURE 6: OPERATION OF A SOLAR PV UNDER PARTIAL SHADING CONDITION [7] ........................ - 10 -
FIGURE 7: BUCK-BOOST CONVERTER [9].......................................................................................... - 12 -
FIGURE 8: ON-STATE OF THE BUCK-BOOST CONVERTER [10] ......................................................... - 12 -
FIGURE 9: OFF-STATE OF THE BUCK-BOOST CONVERTER [10] ........................................................ - 12 -
FIGURE 10: CURRENTS AND VOLTAGES OF THE BUCK-BOOST CONVERTER [9] .............................. - 13 -
FIGURE 11: CIRCUIT DIAGRAM OF SEPIC CONVERTER [11] .............................................................. - 13 -
FIGURE 12: SEPIC WHEN Q1 IS ON [11] ............................................................................................ - 14 -
FIGURE 13: SEPIC WHEN Q1 IS OFF [11]........................................................................................... - 14 -
FIGURE 14: SEPIC COMPONENT VOLTAGES DURING CCM [11] ....................................................... - 15 -
FIGURE 15: SEPIC COMPONENT CURRENTS DURING CCM [11] ....................................................... - 16 -
FIGURE 16: FLYBACK CONVERTER [9]............................................................................................... - 17 -
FIGURE 17: ON-STATE OF THE FLYBACK CONVERTER [12] ............................................................... - 18 -
FIGURE 18: OFF-STATE OF THE FLYBACK CONVERTER [12] .............................................................. - 18 -
FIGURE 19: FRACTIONAL SHORT CIRCUIT CURRENT ALGORITHM [5] .............................................. - 22 -
FIGURE 20: CONTROL-CIRCUIT CONFIGURATION OF THE CONSTANT VOLTAGE TRACKER [5] ........ - 23 -
FIGURE 21: FLOWCHART OF THE ONLY-CURRENT PV METHOD [5] ................................................. - 24 -
FIGURE 22: POS FLOWCHART [5] ..................................................................................................... - 24 -
FIGURE 23: POSSIBLE CASES OF POWER VARIATION FOR THE THREE-POINT WEIGHT COMPARISON
ALGORITHM [5] ....................................................................................................................... - 25 -
FIGURE 24: FLOWCHART OF THE ON-LINE SEARCH ALGORITHM [5] ............................................... - 26 -
FIGURE 25: TOPOLOGY OF A DC-LINK CAPACITOR DROOP CONTROL [5] ........................................ - 26 -
FIGURE 26: THE Β-METHOD FLOW CHART [15] ............................................................................... - 28 -
FIGURE 27: LINEAR APPROXIMATION BETWEEN IMPP AND THE IRRADIANCE LEVEL [5] .................. - 30 -
FIGURE 28: NEURAL NETWORK NET STRUCTURE [7] ....................................................................... - 30 -
FIGURE 29: FLOWCHART FOR THE FA [17] ....................................................................................... - 33 -
FIGURE 30: FLOWCHART OF CONSTANT VOLTAGE METHOD [5] ..................................................... - 34 -
FIGURE 31: FLOWCHART OF THE OPEN-CIRCUIT VOLTAGE METHOD [5] ........................................ - 35 -
FIGURE 32: FLOWCHART OF P&O METHOD [7] ............................................................................... - 36 -
FIGURE 33: FLOWCHART OF INCREMENTAL CONDUCTANCE METHOD [7] ..................................... - 38 -
FIGURE 34: FUZZY RULE TABLE ........................................................................................................ - 39 -
FIGURE 35: MEMBERSHIP FUNCTIONS ............................................................................................ - 39 -
FIGURE 36: CONSTRUCTION OF AN ACCUMULATOR [19] ............................................................... - 40 -
FIGURE 37: EXAMPLE SCHEME OF A PV WITH A MPPT SYSTEM [5] ................................................ - 42 -
FIGURE 38: TEXAS INSTRUMENTS INCORPORATED SOLAR EXPLORER KIT (TOP VIEW) ................... - 43 -
FIGURE 39: SOLAR EXPLORER KIT GUI .............................................................................................. - 43 -
FIGURE 40: BLOCK DIAGRAM EXAMPLE OF BATTERY CHARGE PV SYSTEM [20] ............................. - 44 -
FIGURE 41: CONTROL OF PV STREET LIGHT WITH BATTERY CHARGING [20] .................................. - 44 -
FIGURE 42: DC/DC SEPIC FOR BATTERY CHARGING WITH MPPT [20] ............................................. - 45 -
FIGURE 43: BATTERY CHARGING WITH MPPT CONTROL DIAGRAM [20] ......................................... - 45 -
FIGURE 44: MPPT ALGORITHM ........................................................................................................ - 46 -
FIGURE 45: MPPT ALGORITHM ........................................................................................................ - 47 -
- 105 -
List of Tables
TABLE 1: PROJECT REQUIREMENTS.................................................................................................... - 1 -
TABLE 2: FEATURES OF DIFFERENT CONVERTERS [8] ...................................................................... - 11 -
TABLE 3: RESEARCHED MPPT ALGORITHMS .................................................................................... - 20 -
TABLE 4: COMPARISON OF RECHARGEABLE BATTERIES FOR A MPPT SYSTEM ............................... - 49 -
TABLE 5: TESTING OF DIFFERENT MPPT ALGORITHMS WITH CONVERTER CONCEPT ..................... - 97 -
- 108 -
Appendix
• „Description of tests.pdf“
• “Project schedule.pdf”
• “Requirements list.pdf”
- 109 -
Statutory Declaration
Brest, 19.12.2018