Improved Synchronous Reference Frame Simulation and Experimental Results

Dhurian Rohit Vitoldás

Instituto Superior Técnico

Av. Rovisco Pais, 1049-001 Lisboa, Portugal

E-mail: [email protected]
disadvantages, like a fixed-point operation for power factor
ABSTRACT compensation, they are bulky, and can become resonance.
Other solution is the usage of a diode bridge with more
The proposed method is an adaptation from the “Modified
pulses that takes to a reduction of harmonic content
Synchronous Reference Frame” and it comes with the
introduced into the grid but has some disadvantage like,
willing of suppress the handicap which exists in the classical
needs a big and heavy transformer, unwanted voltage drops
methods when are in presence of a polluted or/with
and the dc-current with a bigger harmonic level.
imbalanced electrical grid, throughout the strategic
The solutions proposed above are static solution it means
positioning of filters.
that if we change the initial parameters to which they were
Therefore the aim proposed to this work is to produce a
projected they become ineffective and may start to degrade
experimental results to do a comparison with the classical
even more the current driven by the grid.
method “Modified Synchronous Reference Frame” with the
So one dynamic method proposed to reduce the harmonic
“Improved Synchronous Reference Frame” having different
content and can also improve the power factor are the Active
grid condition which in this case will be confined between
Filters (AF).
the ideal grid and a polluted grid characterized by 15% of
The AF can be divided in several groups, the Parallel Active
5th harmonic, 1.2% of 7th harmonic and 10% of inverse
Filters (PAF) are also called as Shunt Active Filters, Series
component.
Active Filter (SAF), Universal Active Filter (UAF) among
many others.
Index Terms— Active Filters, Synchronous Reference
In this work is proposed using a Parallel Active Filter
Frame, Dynamic Compensation, Power Filters , Improved
controlled by a strategy based in time-domain, to produce
Synchronous Reference Frame
and study the experimental results achieved with a new
I. INTRODUCTION method based in the widely used Modified Synchronous
Reference Frame Method called Improved Synchronous
With the widespread of the nonlinear elements such as, Reference Frame.
diodes, IGBT’s, MOSFET’s, thyristors, among others, in the
most vulgar applications like ordinary power supplies,
uninterruptible power supply (UPS) etc. , started to diminish
the quality of electricity that is transmitted by the grid due
the introduction of harmonics generated by these nonlinear
elements.
This “poor” quality electrical grid leads us to an early usage
of electrical and mechanical material due the excessive
heating, oscillating torque in electrical rotating machines,
increase of losses and heating in electrical grid lines [2].
To reduce the pollution in the grid it is possible to use Figure 1 Generic schematic of shunt active filters.
several methods to eliminate the harmonic content.
One widely used technique is the usage of L-C filters to a
discrete reduction of selected harmonics usually 5th, 7th, This paper is presented in five parts. Starting with an
11th and 13th and capacitors to improve the power factor of introduction, the following chapter will be presented a
ac loads. resumed explanation of how the Modified Synchronous
These filters have as main advantages it easy Reference Frame works, the third part it will consist in
implementation, low price, although it has several showing the Improved Synchronous Reference Frame,

1
followed by the experimental results achieved and finalizing they have the same frequency of the rotating frame reference
with the conclusions. and all other frequencies will become ripple.

II. MODIFIED SYNCHRONOUS REFERENCE III.IMPROVED SYNCHRONOUS REFERENCE

FRAME FRAME

This method of controlling Active Filters (AF) is based in As it was explained in the previous section on the Modified
time and uses Park Transformation to create the signals Synchronous Reference Frame Method (MSRF), the
which are going to be used to compensate the harmonics reference used to determine the angle, which is going to be
created by the non-linear loads. used by the Park transformation, is the ac three-phase
voltage grid. Therefore, when this grid is polluted with
harmonics or/and unbalance this reference will also appear
distorted.
This distortion in the grid will create an incorrect reference
to the Park transformation leading to an injection of an
inadequate compensation current therefore contributing for
an even higher grid disturbance in that region, so to correct
this situation it is presented the Improved Synchronous
Reference Frame [1].
Consequently, to remove the pollution from the grid signal
be used to create the reference to Park transformation are
applied low pass filters and to remove the unbalance it was
also used a low pass filter in the ∆1 [1].
Figure 2 Modified Synchronous Reference Frame [2]. So, when we use this low pass filters the reference angle
becomes clean, as is shown in the Figure 3.1, which
To determine the angle that will be used by Park compares, experimental results of the transformation angle
transformation in Modified Synchronous Reference Frame obtained with the MSRF and ISRF.
Method (MSRF), it’s used grid voltage to be the reference,
by doing this the rotating reference will be synchronized
with the grid and inherit also it frequency in this case 50 Hz.
The angle will be determined using the following schematic:
MSRF
ISRF

Figure 4 MSRF vs. ISRF Reference angle (Laboratory

Result).

Figure 3 Angle calculation for MSRF [1].

Being γ ul the angle used by the Park transformation, the

angle is very sensitive to grid quality. Therefore if the grid is
polluted and/or unbalanced it will produce a poor reference
and the active filter will not produce accurate compensation
currents. Consequently will contribute for an even higher
degradation of the grid, but if the grid voltage is a three-
phase sinusoidal balanced grid this method shows that is
accurate.
By applying the Park transformation to grid currents, signals
with 50Hz frequency will become dc components, because
1
∆ = eα2 + e β2

2
 ISRF
MSRF

Figure 5 Principle of the proposed method2[2].

As it is expected and shown in Figure 4 when are introduced Figure 6 Pink color is for the Improved Method
filters into the ac main voltage signal, the quality of the used reference, the red color is for the Modified method and
angle is improved becoming sharper but will also delay the the orange color is for the grid current. This test was did
angle, introducing an undesired delay in the current driven with a distortion in the grid of 15% in the 5th harmonic
by the grid. This shifting in the reference angle will provoke
a delay to the system when is creating the signals to
compensate the currents from the non-linear load and when As is possible to see in the previous image and Figure 6 is
charging the DC-Link capacitor, this last fact will have as that the ISRF is delayed comparing with the MSRF of
downside decreasing the power factor and will lead to an almost 36º which is equivalent to 0.002s.
increase consumption of reactive power this happens due the In the following figure is shown in vectorial charts why this
inexistent error control in q coordinate for the current delay occurs.
needed by the capacitor.
In practice is possible to decrease this delay by doing a more
refine tuning to the low pass filter, like changing the cut
frequency, the order of the filter or even changing the type
of filter (e.g. Chebyshev).
If the cut frequency is too high it won’t remove all the non-
fundamental harmonics having the consequence of polluting
the reference but it will lead to a decrease of the delay and if
the cut frequency is diminished it will lead to an even higher
delay, by changing the order of the filter to a lower order it
will need an lower cut frequency because it will take more
time to make the filtering efficiently and even if the filter is
changed the filtering it will not produce much difference
because the order and the cut frequency have a bigger
importance in this compromise.
To simulate these increase of reactive power it will be
connected a resistance of 1,2 kΩ in parallel to the DC-Link
capacitor to simulate a real first harmonic compensation in
the system and verify the delay existent in the referential
created from the filtered ac main voltage Figure 6.
Figure 7 Rotating Reference demonstration of the
method.

It is possible to verify in the previous figure appears three

reference frames the stationary (α, β), and two rotating
reference (d*, q*) and (d’, q’) an angle (θ) and a current
ICapacitor’ .
These vector diagrams are determined with the following
method, the stationary is determined by the Concordia
2
transformation, the reference to the rotating referential (d*,
In this work it wasn’t used a low pass filter in ∆ as is q*) is produced using the grid voltage without any
shown in the figure.

3
“treatment” (MSRF), finally we have the new reference that
we produced using the filtered voltage grid (d’, q’) ISRF. Simulation Experimental
This last reference (d’, q’) appears with a delay of θ due to Line Voltage 50V 50V
the filter that we use to produce a clean reference, which will Grid Frequency 50Hz 50Hz
be used by the several Park transformations needed in the Line Inductance 10mH 10mH
Active Power Filters. Load Resistance 100Ω 100Ω
Here is possible to see why is introduced reactive power into DC-link Capacitor 470µF 470µF
the system when we use the ISRF because when is charged Converter Inductance 10mH 10mH
the capacitor in DC-Link with the Improved referential it Converter Resistance 100mΩ 100mΩ
will produce a Iq_Capacitor* and a Id_capacitor* and these
DC-link Voltage 300V 300V
two currents exists in the (d*, q*) referential so the reactive
Transformer Inductance 0.1mH 0.1mH
power will not be zero as it is in the MSRF but the
advantages are remarkable because it is possible to get DC-Link and a cut frequency for the low pass filter for the
almost half the THD with the ISRF instead using the SRF. Table 1 Laboratory and Simulation Condition

ISRF reference of 150 Hz.

IV. EXPERIMENTAL RESULTS The results in the image follow this order:
The experimental results, were acquired in the laboratory 1-Grid Voltage,
setup provided by the Faculty of Electrical Engineering, 2-Current Driven by the Grid,
belonging to the Warsaw University of Technology. 3-Load Current,
There were collected all experimental results for the α , β 4-Current Injected by the Current.
4th order Butterworth filters used in the ISRF were both
tuned for 150Hz for the results shown in this chapter. The used grid conditions were:
The sampling frequency used is 10 kHz to provide
enough filtering accuracy, because the harmonics produced • Ideal Grid,
by the diode bridge are odd and non multiple of 3 (three) • 15% 5th harmonic,
like are the 1st, 5th, 7th, 11th, 13th and followings. A 13th
harmonic frequency is 650Hz so when is used 10kHz • 15% 5th harmonic and 1.2% 7th harmonic,
sampling frequency it will give 15 samplings in each period, • 15% 5th harmonic, 1.2% 7th harmonic and 10%
so it will be enough to achieve filtering accuracy.[3]
In the laboratory setup it was used dS1103 DSP board unbalance.
to read and acquire the measurements did by the sensors.
One phase schematic view from the laboratory and used
in simulation Figure 8 Schematic view.

Figure 8 Schematic view.

The condition in which the simulation and experimental
results were achieve are in the following table
Figure 9 MSRF Ideal Grid
To compare the methods and verify the results, each method
was submitted to different grid condition using 300 V in

4
 Figure 13 MSRF 15% 5th harmonic and 1.2% 7th
Figure 10 ISRF Ideal Grid harmonic Voltage Grid.

Figure 14 ISRF 15% 5th harmonic and 1.2% 7th

harmonic Voltage Grid.
Figure 11 MSRF 15% 5th harmonic Voltage Grid.

Figure 15 MSRF 15% 5th harmonic, 1.2% 7th harmonic

th
Figure 12 ISRF 15% 5 harmonic Voltage Grid. and 10% unbalance Grid Voltage.

5
 Table 2 - THD results for experimental results using a
th th
Figure 16 ISRF 15% 5 harmonic, 1.2% 7 harmonic DC-Link of 500 V.
and 10% unbalance Grid Voltage.

Table 3 - THD for experimental results using 300 V in

DC-Link.

V. CONCLUSIONS
Figure 17 THD results with 500 V in DC-link.
In this work were presented the experimental results for a
new control strategy based on the Modified Synchronous
Reference Frame Method, called the Improved Synchronous
Reference Frame Method. This new method was conceived
to be insensitive to pollution in the signals which are used to
create the rotating reference frame, in this case was used the
Grid Voltage.
This reference is critical for a good and accurate functioning
of the active filter therefore producing a correct
compensation for the current driven by the grid.
Consequently, to verify the reliability and the improvement
brought by this new method it were created several different
grid conditions using a grid simulator in this case was from
the California Instruments, these created grids went from the
ideal grid condition till the more severe condition, tested in
this work, where the 5th harmonic has 15% from the
Figure 18 THD results with 300 V in DC-Link. fundamental the 7th has 1.2% and also it involves 10% of
unbalance.
Therefore after completing the several grid conditions used
to test the ISRF in the laboratory setup, it is possible to

6
conclude that the Improved Synchronous Reference Frame slow response of the current proportion integral controller
Method has a very good behavior under harsh conditions as due it phase bandwidth limitation, it could be improved by
it is seen in the THD charts provided in chapter IV being using hysteresis controller.
very good.
Is also possible to point out that it, has an easy
implementation in the DSP, provides a much better THD VI. REFERENCES
results as it was shown in Table 2, some times we can reach
[1] G. D. Marques; V. Fernão Pires; Mariusz Malinowski; Marian
50 % less THD (e.g. under very distorted grid condition Kazmierkowski “An Improved Synchronous Reference
such as 10% unbalance 15% of 5th harmonic and 1.2% of Frame Method for Active Filters”.
7th harmonic, Figure.5.18) than the classical method. [2] Mariusz Cichowlas “PWM Rectifier with Activer Filtering”,
Is also possible to conclude that is completely independent 2004
from the harmonic content in the grid voltage as is verified [3] Mariusz Malinowski “Sensorless Control Strategies for
in Table 2 and Table 3 under different grid conditions the Three-Phase PWM Rectifiers”, 2001.
method presented always the same THD, approximately 9%. [4] Gil Marques, “Controlo de Motores Eléctricos”, 2007.
One negative effect resulted with the use of this method is [5] Gil Marques, “Modelo Dinâmico de Máquinas Eléctricas”,
2007.
the introduction of reactive power, created when is used the
angle produced by the improved method this can be
eliminated by creating an error control for the q component
in the 1st harmonic compensation of the DC-Link capacitor.
The reactive power appears from the usage the reference
where this angle is produced it's delayed in comparison with
the load current as is shown the figure.4.

This situation is very clear in the following simulation result:

Figure 19 Grid Current using ISRF (Pink) and Grid

Current Using SRF (Green) and Load Current (Blue)
(Result obtained in simulation).

Because the new reference is delayed and this delay happens

due the introduction of a 1st harmonic to charge the
capacitor, and to compensate the switching losses in the
IGBT’s used by the converter because the reference where is
introduced this 1st harmonic is delayed.
After the insertion of this shift, it will increase the active
power that flows in the system to compensate the harmonics
and therefore it may be necessary to redesign the power
electronic circuit.
Moreover is also verified that with higher voltage in the DC-
link it was achieved better THD to the same shift of grid
current, which is an expected result because after doing a
reading of the laboratorial results is visible that is possible to
improve the inner current loop, because it demonstrates a