A New Single-Phase PLL Structure Based on Second Order

Generalized Integrator

Mihai Ciobotaru, Remus Teodorescu and Frede Blaabjerg

Aalborg University
Institute of Energy Technology
Pontoppidanstraede 101, 9220 Aalborg
DENMARK
mpc~a,iet. aau. dk, ret -_iet. aau. dk, 1ial aau. dk

Abstract - Phase, amplitude and frequency of the utility voltage monitoring is used to ensure that the performances of a grid-
are critical information for the operation of the grid-connected connected system comply with the standard requirements for
inverter systems. In such applications, an accurate and fast operation under common utility distortions as line harmonics
detection of the phase angle, amplitude and frequency of the /notches, voltage sags/swells/loss, frequency variations and
utility voltage is essential to assure the correct generation of the
reference signals and to cope with the new upcoming standards. Phe Jum P .
This paper presents a new phase-locked-loop (PLL) method for The general structure of a single-phase PLL including the
single-phase systems. The novelty consists in generating the grid voltage monitoring is shown in Fig.1. Usually, the main
orthogonal voltage system using a structure based on second difference among divers single-phase PLL methods is the
order generalized integrator (SOGI). The proposed structure orthogonal voltage system generation structure.
has the following advantages: - it has a simple implementation; -
the generated orthogonal system is filtered without delay by the v V, o, ,
same structure due to its resonance at the fundamental
if
frequency, - the proposed structure is not affected by the v v c °
frequency changes. The solutions for the discrete v
implementation of the new proposed structure are also
presented. Experimental results validate the effectiveness of the qv Vd
proposed method.
1. INTRODUCTION T V=

Phase, amplitude and frequency of the utility voltage are

.. ~~~~~~Fig.1. General structure of a single-phase PLL
critical information for the operation of the grid-connected An easy way of generating the orthogonal voltage system
inverter systems. In such applications, an accurate and fast in a single-phase structure is using a transport delay block,
detection of the phase angle, amplitude and frequency of the which is responsible for introducing a phase shift of 90
utility voltage is essential to assure the correct generation of degrees with respect to the fundamental frequency of the
the reference signals and to cope with the new upcoming input signal (grid voltage). A related method, but more
standards. complex of creating a quadrature signal is using the Hilbert
Most recently, there has been an increasing interest in transformation [3]. Another different method of generating
Phase-Locked-Loop (PLL) topologies for grid-connected the orthogonal voltage system is using an inverse Park
systems. The PLL is a grid voltage phase detection structure. Transformation as presented in [1], [3], [4] and [5]. All this
In order to detect this phase an orthogonal voltage system is methods has some shortcomings as follows: frequency
required. In single-phase systems there is less information dependency, high complexity, nonlinearity, poor or none
than in three-phase systems regarding the grid condition, so filtering. Thus, further attention should be paid on single-
more advanced methods should be considered in order to phase PLL systems.
create an orthogonal voltage system [1]-[6]. This paper presents a new method of single-phase PLL
The main task of a PLL structure is to provide a unitary structure based on second order Generalized Integrator (GI).
power factor operation, which involves synchronization of The proposed method is a good alternative for creating an
the inverter output current with the grid voltage, and to give a orthogonal system in single-phase systems compared to
clean sinusoidal current reference. Also using a PLL known method [1]-[5]. This method is further presented and
structure the grid voltage parameters, such as grid voltage experimentally validated.
amplitude and frequency, can be monitored. The grid voltage

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 22,2025 at 08:48:07 UTC from IEEE Xplore. Restrictions apply.
 II. ORTHOGONAL SYSTEM GENERATION Step Response
1 2
The proposed method of creating an orthogonal system is k0.3
depicted in Fig. 2. As output signals, two sine waves (v' and 1 L _-- k-3 l
k=1

qv') with a phase shift of 900 are generated. The component 0~8
V' has the same phase and magnitude as the fundamental of
the input signal (v) [7]. , 06
V E

+ - S
_ v' Kr :~1

qv
x&
I
qv J .................. ....... °0 0.02 0.04 0 06 OM08 0A 0A2
_ _ I. t_rator Time (sec)
~~~~~~~~~~~~~~~(b)
SecondOrder ner~Ii~ed
IJ V en
| w
Fig. 2. General structure of a single-phase PLL
Fig. 3. Bode Plot (a) and Step Response (b) of the close-loop
The presented structure is based on second order tase ucin(d tdfeetvle fgi
generalised integrator (SOGI), which is defined as [7]-[1 1]:
Gl= 2cos(2
s + 2
(1
The tuning of the proposed structure is frequency
dependent, thus problems can occur when grid frequency has
- where co represents the resonance frequency of the SOGI. fluctuations. As a consequence, an adaptive tuning of the
v' structure in respect to its resonance frequency is required.
The closed-loop transfer functions T (Hd dV
L
rsd andd
Therefore, the resonance frequency value of the SOGI is
co -o

qv'
- of the structure presented inFig.
i Fig. 2 are defined as:
adjusted by the provided frequency of the PLL structure.
Hq =-) of the structure presented 2 defmed
The proposed method for creating the orthogonal system
VI kcos has a main advantage compared to known methods (i.e.
Hd (s) = (s) =2 kcs 2 (2) Transport-Delay, Hilbert Transformation, and Inverse Park
v s+kos +0)2 Transformation) [1]-[5]. Only using a simple structure, as it
H (s)- qv (s) ko)2 can be seen from Fig. 2, three main tasks are performed: -
- S2+ kcos + (32 generating the orthogonal voltage system; - filtering the
orthogonal voltage system without delay; - the structure is
-where k affects the bandwidth of the closed-loop system. frequency adaptive.
The Bode representation and the step response of the Using the proposed method the input signal v (grid
waveforms filtered
closed-loop transfer function (Hd = V ) for the proposed
voltage) is resulting two clean orthogonal voltages
v and qv', due to the resonance frequency of the
- f

structure at different values of gain k are shown in Fig. 3(a) SOGI at co (grid frequency). The level of filtering can be set
and (b). from gain k as follows: - if k decreases the bandpass of the
Bode Diagram filter becomes narrower resulting a heavy filtering, but in the
0 same time the dynamic response of the system will become
-20
slower as it can be observed from Fig. 3(b). As it can be seen
from Fig. 3(a), at resonance frequency there is no attenuation
-40 / /------~~ compared to a quite large attenuation outside this frequency.
-60 An example about how it works this method is presented
in Fig. 4. The effect of the filter is depicted with a distorted
-80 grid voltage waveform (Vg) containing notches. The created
-100
90
orthogonal system is represented of v' andqv'. For this
k=O.3 experimental result the gain k was equal to 0.8.
45- -
k=1
k=3
0-- 0 2 3:::
4'--

I. . .. .iS: :' '' '':(a)':

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 22,2025 at 08:48:07 UTC from IEEE Xplore. Restrictions apply.
 50C -8C -
5 Backward Euler method
400 - ------Vg--- qV' - -- -- ------- - Forward Euler method
-85 Trapezoidal method -

200 D -90
- 100 l lll
- -95a- -- -

0)
0 ~~~~~~~~~~~~~~~~~~~~-1 OC
-100 -1 021
(Hz) Frequency
-0 Fig. 5. Phase Bode plots for Forward Euler, Backward Euler and
fm
Trapezoidal methods
-3 --

400 Due to the fact that qv' is not 90 degrees phase shifted

-500 -50C
0.005 0.01 0.015 0.02 0.025 0.03 0.035
~~~~~~~~~~~byv',
a ripple of 100 Hz will appear in the estimated
amplitude and frequency of the input signal as it can be seen
Time (sec) form Fig. 6. The input signal was a clean sinusoid with a
Fig. 4. Distorted grid voltage Vg and generated orthogonal voltage frequency of 50 Hz and 325 units amplitude.
system (V' and its quadrature qV')
330
111. DISCRETISATION OF THE SOGI 3L
.Z 325
The discrete implementation of the orthogonal system 0
E
A
generation structure based on SOGI is described in the <
320 -
following. 0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4
The Euler method is the most used common method in t [s]
order to obtain a Discrete-Time Integrator. The equations of 50.5 - r ref. . . .!.!
this method are presented below: freq
Forward Euler method: 50 *j
y (n) =y (n-1)+fTu(n-i1) W
.5 -.

0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4
For this method, -
IsiS approximated
1

F tm bRy:t[s t [s]
-1 Fig. 6. Estimated amplitude and frequency of the input signal when
TS Z (4) Euler method is used
1-z
Backward Euler method: However, the solution for this inconvenience is to make
y(n) =
y (n-1) u(n) use of more advanced numerical methods for the Discrete-
Time Integrator. Thus, three different methods are described
For this method, - is approximated by: in the following:
s - Trapezoidal method;
- Second order integrator;
1 (5) Third order integrator;
The structure presented in Fig. 2 can be easily A. Trapezoidal method
implemented in a discrete form using the Forward Euler The equation of the integrator using this method is
method for the first integrator (its output isv') and the presentedbelow
Backward Euler method for the second integrator (its output Ts
isqv') in order to avoid an algebraic loop. It is also known y(n) = y(n-l)+ [u(n)+u(n-l)]
that the Discrete-Time Integrator using Euler method does 1
not have an ideal phase of -90 degrees. For this method, - is approximated by:
The phases for the Forward Euler, Backward Euler and s
Trapezoidal methods at different frequencies are shown in Ts I + z-1
Fig. 5. The sampling time (Ts) was set to 1-4. It can be 2 I-z-1 (6)
noticed that at 50 Hz the Forward and Backward Euler As it can be seen from Fig. 5 a phase of-90 degrees can be
methods do not provide a phase of -90 degrees. Therefore, obtained using Trapezoidal method for the whole spectrum
the two outputs ( v' and qv') of the orthogonal system of frequencies. Anyway, the Trapezoidal method can not be
generation structure presented in Fig. 2 will not be exactly in . ..JUSt replacing -1 from Fig. 2 with (6), because an
quadrature. , ~~~~~~~~~~~~~applied s

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 22,2025 at 08:48:07 UTC from IEEE Xplore. Restrictions apply.
 algebraic loop will issue. Therefore, the solution is to use the B. Second order integrator
Trapezoidal method for the close-loop transfer function The equation of the second order integrator is presented
(Hd V) presented in (2) in order to avoid any other below [12]:
algebraic loops. y(n) = y(n-l)+ s [3u(n-1)-u(n-2)]
Replacings byzy z-1 2 1
ic + ,inin (2) will result: For this method,
is - approximated by:
k 2 z-1 TS 3z -2
kT-
-
z12 IZ1 (12)
Ho (z) - Tz+l (7) 2 1-z
(2z-1W2z~2 -1 The implementation of the second order using (12)
C7T +1,J + kcy 1 + co integrator is presented in Fig. 8.
Solving further the equation it results: Out
(2kcoTs)(-)
2 _I z

Ho (z) = 4 (z + (2k)T)(z2 - 1) (z )
In
Fx = 2koTsT
Making the following substitutions 2 and

bringing the equation to a canonical form it will result: 2

Fig. 8. Second order integrator implementation

+x + -x +z-2 C. Third order integrator

Ho (z)_I_x+y+4_ _x+y+4 ((9) The equation of the third order integrator is given in the
2(4-y) 1 (x-Y-4 2 following [12]:
tx+y+4) x+y+4) y(n) = y(n-1)+ T-s 23u (n-1) 16u(n-2) + 5u(n-3)
-

b x 2(4 - y)
t x+i+4 n x + y +4 For this method, - is approximated by:
Substituting b - and x-y-4's
{ x+ y+4 0 2 x+y+4 23z- -16z7 +5z (13)
simple discrete form of (2) is obtained:
y
~~~~12
~~~ 1-f'1
a
Fig. 9 shows the implementation of the third order integrator.
bo +b2z-2
Hd (Z)1 -a2z-2-2(10) +

-alz-l ut
Furthermore, (9) can be represented as follows:
Hd (Z)=bo. -1_1 - /6 5
-2 ( 1)
1alz -a2z
The implementation of the Trapezoidal method using (11)
is depicted in Fig. 7, where w 2TI co.

v 12
Fig. 9. Third order integrator implementation
w qv
+ + _ In Fig. 10, a comparison between the Trapezoidal Method
v (T), second order integrator (2) and third order integrator (3)
is made. As it can be noticed, the best results are obtained
bo a, x a2 x when the third order integrator is used. Anyway, the all three
+ proposed solutions give very good results compared to the
c> ~~~~~~Eulermethod.
Fig. 7. Trapezoidal method implementation

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 22,2025 at 08:48:07 UTC from IEEE Xplore. Restrictions apply.
 500
Fig.325.4- Estimated -tampl T 400 Amp- -i.
e [s]Vg
...ampl3 30
E 325.2-
I I ~~~ I
I I I I ~~~~
I I ~- 200
325.1
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4 (D 10
t [s]
_____________________________~~~~~~~~~~
0 ---
50.01- are u -freq T
-100 1
freq 3
g
<S --ond order inertr Thir orde -Ieraoand Trapezoidal...... w r w t strucureeres-200
U- 49.99- qVa
-V
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4
[s] -5OC
Fig. 10. Estimated amplitude and frequency of the input signal when
Second order integrator, Third order integrator and Trapezoidal Fi.13irimolae ea [s].20.401
.0
Time
of50 Nths k08
00 00
s
01

method are usedFi.1.Givotgsaof5%-Nchsk-08

IV. IV. ~All the experimental results were obtained using the PLL
EXPERIMENTAL RESULTS
EXPERIMENTAL structure presented in Fig. 1. The block "Orthogonal system
An experimental system using a grid simulator (5kVA generation" was replaced with the structure presented in Fig.
AC Power Source - model 5001 ix - Califoria Instruments) 2. For all experimental results presented in this paper, the PI
has been builtin order to test the proposed structure. Thecontroller parameters of the PLL structure were set as
hastrol structure
been b sing A I 03follows: - the settling time Tset=0.06 s; - dumping factor 41.
conto sii
platform.adiinlftes All the measurements have been done without using
The proposed PLL structure based on second order additional filters.
generalized integrator is experimentally validated in the Fig. 11, 12 and 13 show the behaviour of the PLL system
following. based on SOGI under grid voltage sag of 500O. The effect of
400 the gain k (Fig. 2) at two different values is depicted in Fig.
11 and 12 for a grid voltage THD of 10%. It can be noticed
300 .. b that a smaller value of the gain gives a better filtering but
200 AmpI. the dynamic of the system. In Fig. 13 the proposed
PLLThstructure is tested under a high content of notches in the
~~~~~~~~~slows
-300- ---V ----- WV- ---v --- -- 9-1 grid voltage.
Fig. 14 and 15 show a frequency step and sweep response
> ~~~~~~~~~~~~~~from
50Oup to 51 Hz. It can be observed a fast estimation of
-100 the grid frequency. The grid voltage THD was set to 300 for
this experiment.
0 .0 .0 .0 .0 01
qva.1 .1 .102
.1
~~~~~~~~The
(T e-.0.)
PLL behaviour under a phase jump of 60 degrees and
-300 -V ...voltage sag of 2500 is presented in Fig. 16. It can be noticed
~~~~~~~~~~~that
..........

Vg the PLL system repnsaccording toissettling tm

-400 r t i time
0 0.02 0.04 0.06 0.08 0.1
Time [s]
0.12 0.14 0.16 0.18 0.2 (Tset=0.06 s).
Fig. 11. Grid voltage sag of 50% - IO% THD (k=0.48) 51.

4CC~~~~~~~~~~~~~~~~~~~~~5.
300 Ampl. 5
N

200 5.
U-,

250.

-200 49.8.... .... ..freq. ref.

'\ qV 49. --- -- freq.
est.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 22,2025 at 08:48:07 UTC from IEEE Xplore. Restrictions apply.
 VI. REFERENCES
51.6
51.4 [1] L.N. Arruda, S.M. Silva, B.J.C. Filho, "PLL structures
51.2 for utility connected systems", Industry Applications
51 . . 11Conference IAS, 2001, vol.4, pp. 2655 - 2660.
>w 50.8
50.6
[2] N. Saitou, M. Matsui, and T. Shimizu, "A
Strategy of Single-phase Active Filter using a Novel d-q
Control

2 50.4 Transformation", Industry Applications Society IAS,

U_ 50.4 S .2003, pp. 1222-1227.
50.2 [3] S.M. Silva, B.M. Lopes, B.J.C. Filho, R.P. Campana,
50 -------------------
"Performance evaluation of PLL algorithms for single-
49.8 :req-ref phase grid-connected systems", Industry Applications
49.6 est freq Conference, 2004, vol.4, pp. 2259 - 2263.
[4] S.M. Silva, L.N. Arruda, and B.J.C. Filho, , "Wide
Time [s] Bandwidth Single and Three-Phase PLL Structures for
Fig. 15. Grid frequency sweep from 50 to 51 Hz (k-0.8) Utility Connected Systems", 9th. European Conference
on Power Electronics and Applications EPE, 2001, pp.
400 1660-1663.
E200009i ~ ~~~- ----[5] Ciobotaru, R. Teodorescu , F. Blaabjerg,
M. 2:
PLL structures for single-phase gridj4z>
"Improved
inverters", Proc. of
0 ~~~~~~~~~~~~~~PELINCEC
'05, 6 pages, paper ID 106.
C -200 ~~~~~~~~~~~~[6]
M. Ciobotaru, R. Teodorescu, F. Blaabjerg, "Control of
______single-stage single-phase PV inverter", Proc. of EPE '05,
-400C 10 pages, ISBN: 90-75815-08-5.
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65
[7] B. Burger and A. Engler, "Fast signal conditioning in
t 100 r single phase systems" - Proc. of European Conference
° ___ 1 C oon Power Electronics and Applications, 2001.
50 7 --------------[8] X. Yuan, W . Merk, H. Stemmler and J. Allmeling,
"Stationary-Frame Generalized Integrators for Current
O / Control of Active Power Filters with Zero Steady-State
a. Error for Current Harmonics of Concern Under
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 Unbalanced and Distorted Operating Conditions" IEEE
Trans. on Ind. App., Vol. 38, No. 2, 2002, pp. 523 - 532.
X 6 -~ -~ theta ref. [9] R. Teodorescu, F. Blaabjerg, M. Liserre and U. Borup,
est. theta [llD"A New Control Structure for Grid-Connected PV
B Inverters with Zero Steady-State Error and Selective
26. theta re.4 .5 05 05 . .5mto ErHarmonico Compensation",
cieflesbsdo Proc. of APECt'04,
iuodlitra Vol. 1,
12004, pp. 580-586.
a.ipp,nesgeheP srcrbs o [IO] S. Fukuda and T. Yoda, "A novel current-tracking
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 method for active filters based on a sinusoidal intemral
Time[st mode", IEEE Trans. on Ind App., Vol.37, No. 3, 2001,
Fig. 16. Phase jump of 60 degrees and voltage sag ofw25 pp.888 - 895.
[11]sD. N. Zmood and D. G. Holmes, "Stationary Frame
V. CONCLUSIONS V.sCONCLUSIONSeCurrent Regulation Steady-
of PWM Inverters with Zero
~~~~State Error", IEEE Trans. on Power Electronics, Vol.
Ine withisoutper, a new single-phase PLL structurebasedon18, No. 3, May 2003, pp. 814 822.
Iecondtiper aenewasi i se PL sre bsedgon [12] B. Ceangd, C. Nichita, L. Protin and N. A. Cutululis,
senown knowordervecoril
genrais
aproac fo thethre
or is presented.
phse "TheorieBucharest,
sstes, anewTehnicL!i
ne de la commande des systtmes", Editura
2001, ISBN 973- 31-2103-7.
algorithm for single-phase systems was derived. The
proposed structure has the following advantages: - it has a
simple implementation; - the generated orthogonal system is
filtered without delay by the same structure due to its

the proposed method has been validated by the experimental

results.

Authorized licensed use limited to: J.R.D. Tata Memorial Library Indian Institute of Science Bengaluru. Downloaded on February 22,2025 at 08:48:07 UTC from IEEE Xplore. Restrictions apply.