www.ietdl.

org
Published in IET Power Electronics
Received on 2nd October 2013
Revised on 16th December 2013
Accepted on 1st February 2014
doi: 10.1049/iet-pel.2013.0734

ISSN 1755-4535

Reduced junction temperature control during

low-voltage ride-through for single-phase
photovoltaic inverters
Yongheng Yang, Huai Wang, Frede Blaabjerg
Department of Energy Technology, Center of Reliable Power Electronics (CORPE), Aalborg University,
Pontoppidanstraede 101, Aalborg, DK-9220 Denmark
E-mail: yoy@et.aau.dk

Abstract: Future photovoltaic (PV) inverters are expected to comply with more stringent grid codes and reliability requirements,
especially when a high penetration degree is reached, and also to lower the cost of energy. A junction temperature control concept
is proposed in this study for the switching devices in a single-phase PV inverter in order to reduce the junction temperature stress,
and thus to achieve improved reliability of a PV inverter. The thermal stresses of the switching devices are analysed during low-
voltage ride-through operation with different levels of reactive power injection, allowing an optimal design of the proposed
control scheme with controlled mean junction temperature and reduced junction temperature swings. The effectiveness of the
control method in terms of both thermal performance and electrical performance is validated by the simulations and
experiments, respectively. Both test results show that single-phase PV inverters with the proposed control approach not only
can support the grid voltage recovery in low-voltage ride-through operation but also can improve the overall reliability with a
reduced junction temperature.

1 Introduction voltage level (e.g. 0.6 pu), a minimum reactive current

(80% of the rated grid current) should be injected into the
An imperative demand of clean and reliable electricity grid to support voltage recovery. This LVRT requirement
generation from renewable energies (e.g. photovoltaic (PV) tends additionally to include all PV systems that are
and wind energy) has been already in effectiveness in some connected to low-voltage grids, even PV modules [4–6],
countries to get rid of the energy reliance on fossil-based since the PV systems are already on the track to dominate
resources [1, 2]. With the development of advanced power the electricity generation. In fact, there also have been some
electronics technology, the PV systems have been a key grid requirements for a very-high penetration level of PV
technology to realise that demand, and will take a major systems to activate reactive power control in order to solve
part in the electricity generation soon in some areas [3]. voltage rise issues. Hence, the demand of reactive power
Yet, a high penetration degree of PV systems makes the injection will come into reality soon.
grid more decentralised and even more vulnerable. Power electronics systems (e.g. PV inverters), together
Consequently, more stringent and more specific grid with advanced control approaches, could underpin the
requirements are going to be imposed on the future PV performance of future PV systems with the provision of
power systems, especially when a very-high penetration aforementioned ancillary services (e.g. LVRT and reactive
level of PV power systems is coming into reality [2–9]. The power injection) [3–14]. The popularity of transformerless
future PV inverters are expected to provide a full range of PV inverters proves that those topologies can achieve high
functionalities like what the conventional power plants do, efficiency [7, 12, 13], which is always of intense interest in
including ancillary services, such as frequency control order to reduce the cost of energy. Recently, the reliability
through active power control, reactive power controllability has become another important issue for such power
and low-voltage ride-through (LVRT) capability under grid electronics based PV systems operating in a harsh
faults. For instance, the new Italian grid code requires that environment, where the solar irradiance level is not constant
the generation unit with the nominal power exceeding 6 kW and the grid conditions may change suddenly (e.g. voltage
should have LVRT capability, and in the German E.ON sag because of short-circuit faults). As it has been presented
grid code defined for medium- and high-voltage in [15], the PV inverter has been the critical part within the
applications, the PV systems are required to inject reactive most failures of a PV system. Thus, it is quite necessary to
power during ride-through [7–11], which is shown in predict the lifetime of a PV inverter and its components in
Fig. 1. This figure illustrates that, under a certain grid order to improve the entire system reliability [15–22],

2050 IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059

& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-pel.2013.0734
 www.ietdl.org
demonstrated on a 3 kW single-phase full-bridge inverter in
a given operation condition (irradiation level: 1000 W/m2,
ambient temperature: 50 °C) by simulations. Then, the
LVRT operation mode is tested experimentally in a 1 kW
grid-connected system, which can indirectly reflect that the
junction temperature is dependent of the injected reactive
power and the generated active power. Furthermore, a
thermal experimental test is conducted on a single-phase
three-level neutral point clamped (NPC) PV inverter to
directly show the junction temperature controllability
through power regulations. All the test results have shown
that the proposed method offers the possibility to achieve a
reduced or even a constant junction temperature, which is
independent of the operation conditions. The adoption of
Fig. 1 Reactive current injection requirements for the power the proposed control method can help the next generation
systems connected to medium- and/or high-voltage grid defined in advanced PV inverters to fulfil the upcoming grid demands
E.ON grid code [11] in terms of efficiency, stability and reliability.

especially when an accurate knowledge of the mission 2 Reduced junction temperature control
profiles (e.g. ambient temperature and solar irradiance) is
available. 2.1 System description and requirements
A PV inverter typically consists of power switching
devices (e.g. insulated gate bipolar transistors (IGBT) and Since the PV systems are still dominantly for residential
metal-oxide semiconductor field-effect transistor applications at present, single-phase topologies are more
(MOSFET)), where various factors can contribute to the widely used solutions for PV systems [7, 12]. Fig. 2
failures of them, such as thermal stress, electrical stress, represents the hardware schematic of an inductor-capacitor-
mechanical stress, materials of the part and deviation in inductor (LCL)-filter based single-phase single-stage
product process and thus they affect the reliability of the full-bridge PV system with the proposed control structure.
entire inverter [16, 17, 23–26]. Among those factors, the Standard functionalities for a PV inverter are (i)
most commonly observed ones are related to the thermal optimisation/maximisation of the input power, (ii)
stresses, including the mean junction temperature and the manipulation of the inverter output voltage (i.e. voltage
junction temperature swings on the power switching devices control) and (iii) synchronisation with the grid voltage [3,
[15–17]. By means of proper component selection 7, 12]. Hence, in normal operation, the system is required
(considering rated power, advanced packaging technologies, to inject the maximum active power, known as maximum
the most stressed situations and the severe users), effective power point tracking (MPPT) control and typically, the
thermal management and robust design and validation, the MPPT control can be enhanced by integrating a boost stage
reliability of a PV inverter can be improved [27–31]. Thus, between the PV panels and the inverter shown in Fig. 2.
in this paper, an operation mode, which can achieve a Moreover, the injected grid current should have less
reduced junction temperature, is addressed for single-phase harmonic components, which is referred as total harmonic
PV inverter during ride-through operation. This control distortion (THD). In normal operation mode, the PV system
method is based on an appropriate thermal management by is required to operate at a minimum power factor (PF) of
properly allocating the reactive power and the active power 0.85, and the injected current THD should be lower than
in LVRT operation mode. 5% [7, 8]. Traditionally, in response to a grid variation, for
This paper is organised as follows: the viability of the example, a voltage sag or a frequency disturbance, the PV
proposed junction temperature control method is illustrated systems should disconnect from the grid in order to prevent
in § II, including the LVRT requirements and possible from islanding operation [4]. However, with an even higher
reactive power injection strategies. It is followed by some installation rate of PV systems, the disconnection of a
implementation examples. In § III, the analysis is firstly considerable amount of PV systems unintentionally will

Fig. 2 Hardware schematic and control diagram of a single-phase single-stage full-bridge PV system with LVRT capability to achieve a
reduced junction temperature under grid faults based on single-phase PQ theory

IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059 2051

doi: 10.1049/iet-pel.2013.0734 & The Institution of Engineering and Technology 2014
www.ietdl.org
further induce frequency instability (grid variations), leading control strategy. Then, according to Fig. 1, the active
to more serious events, for example, power outage and current (Id) and reactive current (Iq) in the dq-rotating
voltage flickering. reference frame are calculated as
To solve this issue, the grid codes have been updated to
enable LVRT capability for PV systems. Moreover, recent ⎧ 

2
studies have shown the beneficial contributions of LVRT ⎪
⎨ I = I 1 − k2 1 − V
d N g
for PV systems to distributed grid stability, conversion
 (3)
efficiency and reliability [2, 3, 6–9]. To perform LVRT ⎪
⎩ Iq = k 1 − V g IN
functionality, the PV system should have the controllability
to remain connected and also to support the grid voltage
with reactive power injection when a voltage fault is in which Vg is the grid voltage level in pu, k is the slope
confirmed. In other words, the PV system should switch the defined previously in Fig. 1 and it has a minimum value of
mode of operation from MPPT to LVRT by monitoring the 2 pu, and (1 − (1/k)) pu ≤ Vg < 0.9 pu. When a deep voltage
grid voltage amplitude. However, with the reallocation of sag occurs (Vg < (1 − (1/k)) pu), according to Fig. 1, the PV
the injected reactive power and active power during LVRT, system should inject full reactive power to the grid. For
the current stresses as well as the power losses are changed, example, when k = 2 pu and the grid voltage goes below
leading to a redistribution of the thermal stresses on the 0.5 pu, the current should be Id = 0 A and Iq = IN. Based on
power switching devices. Thus, the junction temperature of (3), the corresponding PF in LVRT operation mode can be
the power devices can be manipulated indirectly, which expressed as
offers a possibility to achieve a reduced or constant junction
temperature, and thus also an improved lifetime, which has P P
been an important requirement for power electronics based cos w = = 
PV systems. S P 2 + Q2
⎧ 
The viability of this principle is demonstrated by (1) and
⎪
  
(2), where an example of the IGBT lifetime model is ⎪
⎪ 1− · 1 − Vg
2 1
1 − pu ≤ Vg , 0.9 pu
⎨ k2
presented [31]. k
=  
  ⎪
⎪ 1
⎪
⎩ 0 Vg , 1 − pu

b1 b2  b k
Nf = A · DTj · exp · tON 3 · ib4 (1)
T j mean (4)
 
Tj = f i, v, Ta , fs , ploss , . . . (2) where S is the instantaneous apparent power.
with Nf being the cycle-to-failure numbers, A, β1,2,3,4 being
the coefficients related to the device material, tON is the
switching pulse width and i being the wire current. As it 2.2.2 Constant active power strategy – const.-P: In
has been shown in (1), the lifetime of a power device is this control strategy, the injected reactive current is
significantly affected by the junction temperatures, calculated based on Fig. 1 and the active power is kept to
including the cyclic temperature ΔTj and the mean be the nominal value (Pn). The major purpose of this
temperature Tj_mean. Moreover, it is observed in (2) that the control strategy is to deliver as much energy as possible to
junction temperature Tj is a function of various factors, for the grid, even in the case of a grid voltage sag. Hence,
example current level, i, voltage stress, v, ambient according to the single-phase PQ theory, the current in
temperature, Ta, switching frequency fs and the power dq-rotating reference frame can be given as
losses, ploss. Since the power losses are dependent on the
inverter topology and thus the ratio between the generated ⎧ 1
active power and the injected reactive power, an appropriate ⎪
⎨ Id = IN
allocation of reactive power can possibly contribute to the Vg (5)
⎪

reliability improvement. ⎩I = k 1 − V I
q g N

2.2 Reactive power injection strategies

However, with const.-P control strategy, the currentlevel may

Regarding the reactive power injection strategies for
exceed the limitation of the PV inverter (Imax ≤ Id2 + Iq2 ),
single-phase systems, there are three reactive power control
strategies proposed in [7] – constant peak current strategy, leading to inverter shutdown because of over-current
constant active power strategy and constant active current protection. To prevent the inverter from over-current
control strategy. Those strategies can inject sufficient shutdown, the following condition should be satisfied
reactive power, which is dependent on the voltage sag level,
but with different control objectives. Here, two strategies 

2 I
1
(constant peak current control and constant active power 1 + k 2 · Vg − Vg2 ≤ max (6)
control) are selected in order to illustrate the principle of Vg IN
the proposed method. Both strategies are in compliance
with the grid codes shown in Fig. 1. where (1 − (1/k)) pu ≤ Vg < 0.9 pu, and Imax is the inverter
allowable current level. Similar to constant peak current
2.2.1 Constant peak current strategy – const.-I: The control strategy, when a deep voltage sag happens, the
peak of the injected grid current is kept to be the nominal system should inject full reactive power to the grid, and
(rated) current (IN) during ride-through operation for this meanwhile the active power generation should be deactivated.

2052 IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059

& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-pel.2013.0734
 www.ietdl.org
For this control strategy, the PF can be expressed as

Pn Pn
cos w = = 
S Pn + Q2
2
⎧ 
⎪
2  
⎪
⎪ 1
⎨ 1/ 1 + k 2 · Vg − Vg2 1 − pu ≤ Vg , 0.9pu
k
=  
⎪
⎪ 1
⎪
⎩ 0 Vg , 1 − pu
k
(7)
Fig. 4 Control structure of the proposed junction temperature
where k is defined previously, Pn is the nominal active power control method through active power and reactive power regulation
and S is the apparent power.
Although the const.-P control can output the maximum areas. This is the main idea of the proposed control method
power in different operation modes, it is known from (6) in order to achieve a constant or reduced junction
that the const.-P control strategy can only be adopted when temperature of the power devices.
Imax ≥ 2.236 IN with k being 2.0 pu, which is very large for For example, as it is shown in Fig. 3, a voltage sag (0.3 pu)
the inverter design and will increase the cost of the PV occurs and the const.-I control strategy is firstly activated once
inverter. Moreover, an inappropriate design margin the voltage sag is detected. In this case, the required PF
(e.g. Imax = 1.5 IN) may cause the power devices more should be approximately 0.8 when k = 2 pu. By adjusting k
easily to fail when a grid fault happens. Whereas for the to 3 pu or changing the mode of control to const.- P control
const.-I control, the peak value of the injected grid current strategy, the operation points will change from C to D or
is kept constant during operation, and thus there is no risk from C to A, and thus the injected active power and reactive
of over-current shutdown. power are manipulated, leading to a change of loss
distribution on the power devices. According to (2) and
2.3 Reduced junction temperature control previous discussions, since the power losses have a
significant impact on the thermal behaviour of the power
Based on the previous discussions, the thermal performance devices, a manipulation of the junction temperature is
of a PV inverter is affected by the power losses, which are achieved by varying k or changing the control strategies.
dependent on the grid conditions and the power allocations. The control philosophy of the proposed method has been
Under different control strategies, the ratio between the illustrated in the last paragraph. A detailed implementation
active power and the required reactive power is varied, and of this control strategy is shown in Fig. 4. By optimising
thus the PF. Therefore the junction temperature is affected. the power references (P* and Q*) shown in Fig. 2
According to (4) and (7), the control areas for these two according to voltage sag depths, an allocation of the powers
strategies under grid faults can be plotted in Fig. 3. The can be done and thus the control goals can be achieved. As
upper borders (solid lines) of the control areas (I and II) are it is shown in Fig. 4, the proposed control method can not
the maximum PF (k = 2) that is required under grid faults only meet the reactive power injection requirement but also
according to Fig. 1. Thus, in these control areas (I and II), enhance the reliability performance by means of controlling
the active power and reactive power can be allocated the power device junction temperature through active and
properly and intentionally, and varied by adjusting the slope reactive powers exchanging with the grid. The key of the
k or by changing the control strategies. As aforementioned, implementation of this control method is to find the
this will lead to a redistribution of the power losses on the optimum power references under different grid levels by
switching devices, and thus offers the control possibility to means of look-up tables or mathematical derivations. Thus,
achieve a reduced or constant junction temperature by an appropriate thermal model of the power devices is
optimising the active power and reactive power in those necessary, since the electrical model of the power devices is
coupled with the thermal model through the power losses
on the device [24, 25].
Fig. 5 shows a thermal model of the power devices, where
Z th is the thermal impedance between two nodes. It is
illustrated in Fig. 5 that the power losses (Ptot(S) and Ptot(D))
on the power devices will cause temperature rise at different
nodes because of the thermal impedances. This relationship
can be described as

Tj(S/D) (t) = Ptot(S/D) (t)Zth(S/D)(j−c) (t) + Tc (t)

= Ptot(S/D) (t)Zth(S/D)(j−c) (t)

   
+ PtotS (t) + PtotD (t) · Zth(c−h) (t) + Zth(h−a) (t)

+ Ta (t)
Fig. 3 PF curves against voltage levels for different control (8)
strategies according to Fig. 1: solid lines: k = 2.0 and dashed
lines: k = 3.0 in which, Tj(S/D) is the IGBT/diode junction temperature,

IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059 2053

doi: 10.1049/iet-pel.2013.0734 & The Institution of Engineering and Technology 2014
www.ietdl.org

Fig. 5 Thermal models from a full-bridge PV inverter in Fig. 2 to a single power device and the thermal impedance [26, 32–34]
Notes: Tj – junction temperature, Tc – case temperature, Th – heatsink temperature, Ta – ambient temperature.

Ptot(S/D) is the IGBT/diode total losses, Zth(S/D)( j-c) is dependent on the voltage sag level. One method is based on
the thermal impedance from junction to case, Zth(c-h) is the mathematical derivations, and it requires a better knowledge
thermal impedance from case to heat sink, Zth(h-a) is the of the power device materials, the topologies, the switch
thermal impedance from heat sink to ambient, Tc is the case schemes etc. Meanwhile, the coupled relationship between
temperature, and S represents the IGBT and D denotes the power losses and the junction temperature further increase
diode. Typically, the thermal impedance can be modelled as the derivation complexity. An alternative is based on
a multi-layer Foster model [28, 33–35], which is a look-up tables, although at a cost of the accuracy. This
series-connected model as shown in Fig. 5, and it can be method is simple and can easily be implemented. Thus, in
expressed as this paper, the look-up table based implementation method
is chosen for a single-phase PV system.

4 To create a satisfied look-up table and thus to find the
Zth(S/D) (t) = Rthi (1 − e−t/ti ) (9) optimum power references for the proposed method,
i=1 different cases (varying allocations of reactive power and
active power) should be studied first. Referring to Fig. 2, a
with ti = CiRthi. Those parameters in (9) can be found in the 3 kW single-phase system is simulated under various active
datasheet. It is shown in (8) and (9) that the junction power levels with reactive power injection when a
temperature is dependent on the dissipated power. This also low-voltage fault occurs. The results are presented in Fig. 6,
implies the viability of the proposed method. where it is noted that the maximum junction temperature
(Tjmax) of the power devices will exceed the allowable
2.4 Implementation of the proposed method value (e.g. 125 °C) under very-low-voltage conditions (e.g.
(power references) 0.4 pu). It also reveals that the maximum junction
temperature will increase with injecting required reactive
Regarding the implementation of the proposed method, there power in const.-P control mode. By decreasing the active
are two possibilities to obtain the power references, which are power output and limiting the reactive power injection, the

Fig. 6 Maximum junction temperature (Tjmax) of a full-bridge inverter (Pn = 3 kW) under grid faults with different reactive power injection
a P = 80% Pn
b P = 60% Pn
c P = 40% Pn
d P = 20% Pn

2054 IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059

& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-pel.2013.0734
 www.ietdl.org

Fig. 7 Reactive power and active power references for a PV inverter

a To comply with the grid requirement of reactive power injection during LVRT, ‘L’ – LVRT, k = 2 pu (grey) and k = 3 pu (black) in respect to Fig. 1
b To achieve a constant junction temperature (Tjmax_d = 80 °C), ‘J’ – Junction and
c Flowchart of the proposed junction temperature control method

maximum junction temperature could be controlled below a is shown in Fig. 7c. It should be pointed out that, by
desirable value (e.g. Tjmax_d = 80°C) and thus can be kept applying curve-fitting to Fig. 7b, the power references can
almost constant during ride-through. Considering the be formulised.
required reactive power injection in Fig. 1, the active power
and reactive power references can be obtained as shown in 3 Simulation and experimental results
Fig. 7a under different voltage sag levels. Whereas, for a
constant junction temperature of 80°C, based on Fig. 6, the 3.1 Simulation tests
power references can be obtained and shown in Fig. 7b.
Consequently, the reference generation zones can be As it is discussed above, by adjusting the active power and
implemented as the ‘power reference generation’ unit in reactive power according to Fig. 7 and (10), a reduced or
Fig. 2. In the detailed control structure of the proposed even constant junction temperature operation can be
method in Fig. 4, the power references for LVRT achieved for single-phase grid-connected PV inverter. The
requirements (Fig. 7a) and constant junction temperature hardware and overall control structure of the single-phase
(Fig. 7b) can be implemented as the ‘grid requirements’ and PV system is shown in Fig. 2. Instead of PV panels, a DC
‘thermal optimisation’ units, respectively. source is used in this system on the assumption that the
For instance, when the grid sags to be 0.6 pu, there are MPPT control is very robust. The proposed control method
several sets of power references available in Fig. 7b – (i) shown in Fig. 7 was implemented in MATLAB/Simulink.
P*J = 0.2 pu and Q*J = 0.68 pu; (ii) P*J = 0 pu and Q*J = A 0.43 pu voltage sag is generated in the simulations. The
0.68 pu; and (iii) P*J = 0 pu and Q*J = 1.0 pu. For the first second-order generalised integrator based phase locked loop
case, the active power should be 0.2 pu ( point B in [7] is adopted to synchronise the current with the grid
Fig. 7a). One way to generate this active power is to voltage. In order to directly and intuitively control the
change the slope k, being 2.36 pu, while the corresponding active power and reactive power, the single-phase PQ
reactive power Q*L = 0.57 pu ( point D in Fig. 7a). This control method is adopted in this system. A proportional
reactive power is insufficient to keep the junction resonant current controller with harmonic compensators has
temperature constant as shown in Fig. 7b ( point A), and been selected in order to guarantee a good power quality.
thus, the maximum junction will be lower than Tjmax_d. If An IGBT module (600 V and 50 A) from a leading
the required reactive power (Q*J = Q*L = 0.68 pu) is manufacturer is selected with the thermal parameters shown
injected to the grid, the peak value of the injected current in Table 1. The other specifications of the system are listed
will not be constant during LVRT, however, the maximum in Table 2.
junction temperature will be kept constant. Another way to The power losses and the cyclic temperature on the power
achieve a constant junction temperature is to further devices of a 3 kW single-phase PV system are firstly tested
decrease the active power generation either by increasing k for constant junction temperature control. When a voltage
or chaning the reactive power injection strategies (from A to fault (0.43 pu) is detected, the system can immediately be
C), as it is shown in Fig. 7. changed to the proposed control operation mode from
According to Fig. 6, the junction temperature of the power normal operation condition with the MPPT control. For
devices of a full-bridge inverter is proportional to the injected
reactive power and the generated active power. Thus, to
ensure a reduced junction temperature and considering the Table 1 Foster thermal parameters for an IGBT module from a
leading manufacturer
grid requirements, the power references optimised by the
central control unit in Fig. 4 can be given as Impedance Zth( j-c) Zth(c-h)

   I 1 2 3 4
P∗ = min PL∗ , PJ∗
  (10) IGBT Rthi (K/W) 0.074 0.173 0.526 0.527 0.7
Q∗ = min Q∗L , Q∗J ti (s) 0.0005 0.005 0.05 0.2 0
diode Rthi (K/W) 0.123 0.264 0.594 0.468 0.7
ti (s) 0.0005 0.005 0.05 0.2 0
Consequently, the control flowchart of the proposed method

IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059 2055

doi: 10.1049/iet-pel.2013.0734 & The Institution of Engineering and Technology 2014
www.ietdl.org
Table 2 Simulation parameters for a single-phase the results presented in Table 3 and Fig. 8a. By applying
grid-connected PV system the proposed junction temperature control, the power losses
nominal grid voltage VgRMS = 230 V on the switching devices of a single-phase PV system are
nominal grid frequency f = 50 Hz significantly reduced in LVRT operation mode, as it is
rated power Pn = 3 kW shown in Table 3. The resultant lower power losses lead to
nominal DC voltage (PV output Vdc = 400 V a reduced mean/maximum junction temperature compared
voltage)
switching frequency fsw = 10 kHz
with the case in const.-P mode during ride-through, as it is
LCL filter Lif = 3.6 mH, Cf = 2.35 μF, proved in Fig. 8a. Moreover, the IGBT temperature cycling
Lig = 4 mH amplitude is also reduced to 10°C (ΔTj2) from 30°C (ΔTj1),
grid impedance Lg = 50 μH, Rg = 0.1 Ω when the system is controlled in const.-P mode as it is
shown in Fig. 8a. Thus, with the proposed control method,
a constant maximum junction temperature is achieved
comparison, the LVRT operation mode with const.-P and during ride-through with sufficient reactive power injection,
const.-I control strategies are also simulated. The results are and thus the overall reliability of the PV system is further
shown in Table 3 and Fig. 8. improved according to (1). However, it can be noted from
As it can be observed in Fig. 8a, although the LVRT Fig. 8a that during ride-through the diode junction
operation with const.-P control strategy can also inject the temperature is increased because of the injection of reactive
required reactive power, the injected grid current level is power into the grid. The diode junction temperature might
about 2.0 pu during LVRT. The high current amplitude will also be reduced or by applying the constant junction
lead to a higher power losses and thus according to (2) the temperature control. In that case, the control objectives of
mean/maximum junction temperature will increase the entire system during LVRT will be: (i) to ride-through
significantly in LVRT operation, which has been verified by low-voltage condition; (ii) to keep the IGBT junction

Table 3 Power loss distributions of the IGBT modules of a 3 kW single-phase PV inverters in different operation modes (unit: W)
Module no. 1 2 3 4

S1 D1 S2 D2 S3 D3 S4 D4

normal operation (MPPT) 19.2 3.1 18.95 3.1 18.95 3.1 19.2 3.1
low-voltage ride-through (const.-P, 0.43 pu) 48.1 13.6 47.65 13.47 47.65 13.47 48.1 13.6
constant junction temperature control (0.43 pu) 18.7 8.1 18.4 8.0 18.4 8.0 18.7 8.1

Fig. 8 Simulation results of a 3 kW single-phase full-bridge PV inverter with the proposed control method to
a Achieve constant maximum junction temperature
b Reduce junction temperature (voltage sag depth: 0.43 pu): ig: grid current, vg: grid voltage, P: injected active power to the grid, Q: injected reactive power to the
grid, Tj: junction temperature, S: IGBT, D: diode

2056 IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059

& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-pel.2013.0734
 www.ietdl.org
temperature constant; and (iii) to reduce the thermal stress on specifications of the simulation system listed in Table 2. The
the diode. Similar power reference profiles (Fig. 7) can be experimental results are presented in Fig. 9.
obtained in order to achieve these goals. During the LVRT operation shown in Fig. 9, the
Fig. 8b shows the results of a 3 kW system with the single-phase system is injecting reactive power into the grid
proposed control method to achieve a reduced junction according to grid requirements defined in Fig. 1. At the
temperature during LVRT. By adjusting the slope k, the same time, the active power is reduced in order to achieve a
maximum junction temperature is reduced during LVRT. reduced junction temperature of the IGBT devices. When
Compared with the constant junction temperature control the voltage fault is cleared (the voltage amplitude goes to
and the const.-I control strategy, the proposed control 90% of the nominal value), the system returns to its normal
method can fulfil the requirement of power injection operation mode and it is injecting satisfactory current at
requirement and also improve the reliability of the entire unity PF. Since the const.-I control strategy are adopted in
system, since the power references for this control method the experiments, this experimental test shows that, by
are optimised according to Fig. 7 and (10). Those results injecting reactive power, a constant amplitude of the grid
show the effectiveness of the proposed method to achieve a current is achieved, leading to a reduced junction
reduced or even a constant junction temperature of the temperature as discussed above. It also demonstrates the
power devices under grid faults. flexibility of a single-phase system to provide
multi-functionalities in the future. The single-phase PQ
power control method in the test is effective in terms of fast
3.2 Experimental verifications dynamic response and reduced thermal stress. Furthermore,
the dynamic behaviours of the active power and reactive
In order to demonstrate the ability of reactive power injection power shown in Fig. 9b are similar to those in Fig. 8 during
under grid faults for future PV inverters, a 1 kW the voltage recovery period (i.e. voltage sag is cleared),
grid-connected system is examined in the laboratory based which shows the effectiveness of the proposed junction
on single-phase PQ theory [7, 30]. The sag generator is used temperature control method.
to make a 0.43 pu voltage sag in the experiments. The Since the lack of open IGBT modules and thus the
proportional resonant controller with harmonic compensators corresponding thermal testing setup for the 3 kW
is adopted again to achieve high power quality of the single-phase PV inverter, of which many efforts will be
injected current. Except for the grid impedance, the other devoted to in the future, the illustration of the junction
parameters of the experimental setup are the same as the temperature controllability is experimentally demonstrated

Fig. 9 Experimental results of a 1 kW single-phase grid-connected system in LVRT operation modes (0.43 pu voltage sag): time [40 ms/div]
a Grid voltage vg [100 V/div] and grid current ig [5 A/div]
b Active power P [500 W/div] and reactive power Q [500 Var/div]

Fig. 10 Experimental results (thermal performance) of a single-phase three level NPC PV inverter under different PFs: Sp1 – IGBT, Sp2 –
clamped diode and Sp3 – MOSFET

IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059 2057

doi: 10.1049/iet-pel.2013.0734 & The Institution of Engineering and Technology 2014
www.ietdl.org
Table 4 Test results of a single-phase three-level NPC PV 4 Carnieletto, R., Brandão, D.I., Suryanarayanan, S., Farret, F.A., Simões,
inverter under different PFs M.G.: ‘Smart grid initiative –a multifunctional single-phase
voltage-source inverter’, IEEE Ind. Appl. Mag., 2011, 17, (5), pp. 27–35
Test Power P Q S Hotspot 5 Xue, Y., Divya, K.C., Griepentrog, G., Liviu, M., Suresh, S., Manjrekar,
No. factor (W) (Var) (VA) temperature M.: ‘Towards next generation photovoltaic inverters’. Proc. ECCE 2011,
Phoenix, Arizona, September 2011, pp. 2467–2474
Sp1 Sp2 Sp3 6 Papanikolaou, N.P.: ‘Low-voltage ride-through concept in flyback
inverter-based alternating current-photovoltaic modules’, IET Power
1 0.4538 257.4 505.5 567.3 23.7 23.3 22.6 Electron., 2013, 6, (7), pp. 1436–1448
2 0.8955 507.5 252.3 566.8 24.1 24.6 23.6 7 Yang, Y., Blaabjerg, F., Wang, H.: ‘Low voltage ride-through of
3 0.9996 564.7 16.8 565 23.7 24.9 23.4 single-phase transformerless photovoltaic inverters’, IEEE Trans. Ind.
Appl., 2014, early access at: http://www.dx.doi.org/10.1109/TIA.2013
.2282966
8 Comitato Elettrotecnico Italiano, CEI 0–21: ‘Reference technical rules
on a three-level NPC high power PV inverter operating at low for connecting users to the active and passive LV distribution
companies of electricity’, December 2011
power conditions. A commercial three-level NPC PV inverter 9 Bae, Y., Vu, T.-K., Kim, R.-Y.: ‘Implemental control strategy for grid
with the rated current of 30 A and the rated voltage of 1200 V stabilization of grid-connected PV system based on german grid code
is selected as the candidate for thermal tests, and each leg of in symmetrical low-to-medium voltage network’, IEEE Trans. Energy
the NPC inverter consists of two IGBTs, two clamped diodes Convers., 2013, 28, (3), pp. 619–631
and two MOSFETs. The switching frequency is 20 kHz and 10 El Moursi, M.S., Xiao, W., Kirtley, J.L.: ‘Fault ride through capability
for grid interfacing large scale PV power plants’, IET Gener. Transm.
the experimental results are shown in Fig. 10 and Table 4. Distrib., 2013, 7, (9), pp. 1027–1036
As it is observed in the test results, the junction temperature 11 Benz, C.H., Franke, W.-T., Fuchs, F.W.: ‘Low voltage ride through
of the clamped diode (i.e. hotspot Sp2) experienced a linear capability of a 5 kW grid-tied solar inverter’. Proc. EPE-PEMC 2010,
change with the PF, which is in coincidence with Ohrid, Republic of Macedonia, September 2010, pp. T12-13–T12-20
12 Kjaer, S.B., Pedersen, J.K., Blaabjerg, F.: ‘A review of single-phase
discussions presented in [28]. While the junction grid-connected inverters for photovoltaic modules’, IEEE Trans. Ind.
temperature of the IGBT is kept the same, however, the PFs Appl., 2005, 41, (5), pp. 1292–1306
are different (test nos 1 and 3). This verifies the possibility 13 Meneses, D., Blaabjerg, F., García, O., Cobos, J.A.: ‘Review and
to achieve a constant (or reduced) junction temperature of comparison of step-up transformerless topologies for photovoltaic
the power devices by appropriately allocating the active AC-module application’, IEEE Trans. Power Electron., 2013, 28, (6),
pp. 2649–2663
power and reactive power, which is the essential idea of this 14 Miret, J., Camacho, A., Castilla, M., Garcia de Vicuna, J., Matas, J.:
paper. ‘Control scheme with voltage support capability for distributed
generation inverters under voltage sags’, IEEE Trans. Power
Electron., 2013, 28, (11), pp. 5252–5262
15 Collins, E., Dvorack, M., Mahn, J., Mundt, M., Quintana, M.:
4 Conclusions ‘Reliability and availability analysis of a fielded photovoltaic system’.
Proc. PVSC 2009, Philadelphia, Pennsylvania, June 2009,
An approach to achieve a reduced or constant junction pp. 2316–2321
temperature of the switching devices in a single-phase PV 16 Wang, H., Liserre, M., Blaabjerg, F.: ‘Toward reliable power electronics:
challenges, design tools, and opportunities’, IEEE Ind. Electron. Mag.,
inverter has been proposed in this paper. The controlled 2013, 7, (2), pp. 17–26
constant junction temperature and reduced temperature 17 Huang, H., Mawby, P.A.: ‘A lifetime estimation technique for voltage
swings allow a further improved reliability of the PV source inverters’, IEEE Trans. Power Electron., 2013, 28, (8),
inverter and an improved robustness of the PV system in pp. 4113–4119
18 Sujod, M.Z., Erlich, I., Engelhardt, S.: ‘Improving the reactive power
response to abnormal grid conditions (e.g. LVRT). By capability of the DFIG-based wind turbine during operation around
appropriately injecting reactive power to the grid during the synchronous speed’, IEEE. Trans. Energy Convers., 2013, 28, (3),
ride-through operation, the single-phase PV inverter would pp. 736–745
not be at the risk of over-current protection and therefore 19 Muray, C.J., Davoudi, A., Chapman, P.L.: ‘Reliability analysis for
with a better grid support capability. Simulation results on a single-phase photovoltaic inverters with reactive power support’. Proc.
PECI 2011, Champaign, Illinois, February 2011, pp. 1–6
3 kW PV inverter with different levels of reactive power 20 Petrone, G., Spagnuolo, G., Teodorescu, R., Veerachary, M., Vitelli, M.:
injection and experimental results from a 1 kW prototype ‘Reliability issues in photovoltaic power processing systems’, IEEE
verified the effectiveness of the presented control scheme. Trans. Ind. Electron., 2008, 55, (7), pp. 2569–2580
The proposed control method has been also demonstrate on 21 Harb, S., Balog, R.S.: ‘Reliability of candidate photovoltaic
a three-level NPC PV inverter, and the results confirmed the module-integrated-inverter (PV-MII) topologies – a usage model
approach’, IEEE Trans. Power Electron., 2013, 28, (6), pp. 3019–3027
controllability of the junction temperature through the 22 De Leon-Aldaco, S.E., Calleja, H., Chan, F., Jimenez-Grajales, H.R.:
injected power regulation. Considering that the voltage fault ‘Effect of the mission profile on the reliability of a power converter
is a short period, the proposed method also offers the aimed at photovoltaic applications – a case study’, IEEE Trans. Power
possibility to maintain a constant junction temperature of Electron., 2013, 28, (6), pp. 2998–3007
switching devices under varying solar irradiance levels if a 23 Hitachi, Ltd.: ‘High Voltage IGBT Module Application Manual’,
December 2009, pp. 1–33
wide range of reactive power injection is allowed by future 24 Dieckerhoff, S., Bernet, S., Krug, D.: ‘Power loss-oriented evaluation of
grid codes. high voltage IGBTs and multilevel converters in transformerless traction
applications’, IEEE Trans. Power Electron., 2005, 20, (6),
pp. 1328–1336
25 Dupont, L., Avenas, Y., Jeannin, P.-O.: ‘Comparison of junction
5 References temperature evaluations in a power IGBT module using an IR camera
and three thermosensitive electrical parameters’, IEEE Trans. Ind.
1 REN 21: ‘Renewables 2013: Global Status Report (GSR)’, available at: Appl., 2013, 49, (4), pp. 1599–1608
http://www.ren21.net/, June 2013 26 Ji, B., Pickert, V., Cao, W., Zahawi, B.: ‘In Situ diagnostics and
2 Kovanen, K.O.: ‘Photovoltaics and power distribution’, Renew. Energy prognostics of wire bonding faults in IGBT modules for electric
Focus, 2013, 14, (3), pp. 20–21 vehicle drives’, IEEE Trans. Power Electron., 2013, 28, (12),
3 Malashenko, E., Appert, S., al-Mukdad, W.: ‘Advanced inverter pp. 5568–5577
technologies report’ (California Public Utilities Commission, 2013), 27 Ma, K., Blaabjerg, F.: ‘Modulation methods for neutral-point-clamped
pp. 1–13 wind power converter achieving loss and thermal redistribution under

2058 IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059

& The Institution of Engineering and Technology 2014 doi: 10.1049/iet-pel.2013.0734
 www.ietdl.org
low-voltage ride-through’, IEEE Trans. Ind. Electron., 2014, 61, (2), 31 Wagenitz, D., Hambrecht, A., Dieckerhoff, S.: ‘Lifetime evaluation
pp. 835–845 of IGBT power modules applying a nonlinear saturation
28 Ma, K., Liserre, M., Blaabjerg, F.: ‘Reactive power influence on the voltage observer’. Proc. CIPS 2012, Nuremberg, Germany, March
thermal cycling of multi-MW wind power inverter’, IEEE Trans. Ind. 2012, pp. 1–5
Appl., 2013, 49, (2), pp. 922–930 32 Lutz, J., Schlangenotto, H., Scheuermann, U., De Doncker, R.:
29 Murdock, D.A., Ramos Torres, J.E., Connors, J.J., Lorenz, R.D.: ‘Active ‘Semiconductor power devices: physics, characteristics, reliability’
thermal control of power electronics modules’, IEEE Trans. Ind. Appl., (Springer, 2011)
2006, 42, (2), pp. 552–558 33 ABB: ‘Application Note: Applying IGBTs’, May 2012, pp. 1–40
30 Yang, Y., Wang, H., Blaabjerg, F., Ma, K.: ‘Mission profile based 34 ON Semiconductor: ‘IGBT Applications (Handbook)’, July 2012
multi-disciplinary analysis of power modules in single-phase 35 Plexim: ‘The simulation platform for power electronic systems (User
transformerless photovoltaic inverters’. Proc. EPE’13 ECCE Europe Manual Version 3.4)’, http://www.plexim.com/download/
2013, Lille, France, September 2013, pp. P.1–P.10 documentation, June 2013

IET Power Electron., 2014, Vol. 7, Iss. 8, pp. 2050–2059 2059

doi: 10.1049/iet-pel.2013.0734 & The Institution of Engineering and Technology 2014