This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics.

This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

An Advanced Switching Harmonic Cancellation

Method for A Dual-Generator Power System in
More-Electric Aircraft
Cheng Wang, Tao Yang, Senior Member, IEEE, Xiaoyu Lang, Serhiy Bozhko, Senior Member, IEEE

Abstract- Dual-Generator power systems have attracted Idc DC component of the current idc on DC-bus.
increased attention in recent years as it can significantly Idc,i,j Magnitude of DC-side current harmonics in the
reduce the fuel assumption of aircraft engines allowing frequency of ifc +jf0
power transfer between different engine shafts. In such a Ihf Total harmonics on DC capacitor
system, two generators supply a common DC bus together.
A DC-bus capacitor is required to filter switching k k=0, 1 and 2 represent phase A, B and C respectively
harmonics to ensure that the DC-bus voltage satisfying Km,n Harmonic amplitude of switching function
MIL-STD-704F standard. Due to the high current rating,
Ld d-axis inductance
this capacitor is normally bulky and heavy. This paper
aims to introduce a switching harmonic cancellation Lq q-axis inductance
method which can effectively reduce harmonics at the DC P Converter power
bus. This, in return, will extend the lifetime and reduce the
weight of the dc-bus capacitor. With the assumption that R Stator resistance
power converters connected to the DC bus are modulated sf(t) Switching function of the AC-DC converter
with the commonly used space vector pulse width ΔT Phase-shift time
modulation (SVPWM) technique, a simplified
mathematical model to estimate both first and second Ts Switching cycle
switching harmonics is developed. Based on the proposed Vac Magnitude of converter phase voltage
models, an advanced method of switching harmonic vd d-axis stator voltage
cancellation scheme is proposed through an adaptive phase
shift of carrier signals for different power output of vq q-axis stator voltage
converters. Both simulation and experimental results are α Power factor angle
used to validate the proposed model and the suppression
method. β Angle between phase current and its AC side voltage
θc Angle of the carrier signal
f0 Fundamental frequency from PMSG.
Δθc Phase-shift angle between the two AC-DC converters
fc Switching frequency
θc [k]
Phase angle of the triangular carrier signal for each
i, j Switching and band side orders of DC-side current leg
harmonics θc{1} Phase angle of carrier signals on converter {1}
id d-axis current
θc{2} Phase angle of carrier signals on converter {2}
iq q-axis current [𝑘] Phase angles of components in switching function
𝜎𝑚,𝑛 ,
idc DC-side current [𝑘]
𝜑𝑚,𝑛
idc,i,j (t) DC-side current in the frequency of ifc +jf0 φm Flux linkage of the permanent magnet
I1{1} Magnitude of the first DC-side switching harmonic ωe Electrical rotor angular velocity
from Converter 1
I1{2} Magnitude of the first DC-side switching harmonic
from Converter 2
I2{1} Magnitude of the second DC-side switching I. INTRODUCTION
harmonic from Converter 1 More electric aircraft (MEA) concept has been one of the
I2{2} Magnitude of the second DC-side switching major trends towards future aerospace development due to
harmonic from Converter 2 decreased fuel consumption, low maintenance cost, etc. [1], [2].
Iac Amplitude of the fundamental component of AC On board MEA, pneumatic, hydraulic, and mechanical
current subsystems are replaced by their electrical equivalences, which

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912
IEEE Journal of Emerging and Selected Topics in Power Electronics
result in a significant increase of power demand. To give more
sufficient and reliable power supply, single DC-bus power
system with parallel connected generators has attracted
significant interest [3]. With such structure, expected benefits
include less need of cables, convenient integration of energy
storage system, and higher redundancy ability under fault
condition. Meanwhile, multi-generator power system are being
widely considered not only for MEA, but also for hybrid
vehicle [4], ships [5] and microgrid [6].
With the increase of electrical load on board, more electrical
power needs to be extracted from aircraft engines.
Conventionally, electrical power is extracted from
high-pressure spool of an aircraft engine. Nowadays, the engine
manufacturers are trying to further improve the engine
efficiency through increasing the bypass ratio (ratio of air
passing through the bypass over air passing through the engine
core). The increase of the bypass ratio will limit the power
extraction from the high-pressure spool. This makes electrical
power extraction from the low-pressure spool inevitable. As
shown in Fig. 1 from [7], two electrical generators are driven by
two engine shafts, i.e., high pressure (HP) shaft and low
pressure (LP) shaft, respectively. There, authors have
concluded that this architecture enables dual-channel power
generation for more-electric aircraft application. This structure
also enables power transferring between the two engine shafts
and improve engine efficiency and stability margin, as well as
compatible thrust with flight mission [3].
This paper is to further optimize the system shown in Fig. 1
and minimize the capacitance on DC bus while maintaining the
power quality. Capacitors are always bulky and expensive.
Minimizing weight and volume of capacitor is critical which
can potentially reduce fuel consumption and improve
maintainability. Moreover, the capacitor lifetime will be
extended when its voltage across is with lower harmonics [8].
Active suppression of harmonics across capacitors has received
significant attention recently and related studies have been
reported in many publications.
+ a mathematical way to estimate the magnitudes and the phase
AC/DC
- angles of the harmonics.
HP
PMSG1
Capacitor
AC/DC
PMSG2
Resistor load
LP
Fig. 1 Dual-channel enhanced power generation system.
In [9], voltage harmonics across a capacitor are reduced with
a variable DC bus voltage. In [10] and [11], researchers
investigated minimizing harmonics when a Boost converter is
connected with 2-level DC-AC converter. These methods only
discuss application with single 2-level AC-DC converter,
which is different from dual-generator system in this paper.
Adding additional circuit is another solution for reducing
capacitor harmonics [12][13]. There, an additional switching
© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
device were added to the main circuit and voltage harmonics on
capacitors were effectively reduced under desired power flow
with adjustable switching operation of the additional switches.
However, this method requires extra elements and thus
increases the cost and complexity.
In [14] and [15], Zhang investigated the methods of reducing
the second-order harmonic on DC-link harmonics, which is
under low frequency range. In [16] and [17], harmonics on a
DC capacitor were reduced on paralleled three-phase
voltage-source converters through a fixed phase shift angle on
converters. However, the analysis is based on the assumption
that power converters are sharing a common AC fundamental
frequency, which is not the case of dual-generator power
system.
For the dual-generator power system, our previous work has
demonstrated a harmonic cancellation method when the power
converters are with SPWM modulation scheme [18]. The
second switching harmonic is suppressed specifically. Our
paper extends the previous work and develop a harmonic
cancellation method for SVPWM modulation. Compare with
SPWM, SVPWM gives higher modulation index limitation, but
the harmonic distribution of DC-link current is more
complicated. For instance, there is a harmonic component on
the first switching frequency when using SVPWM. However,
no harmonic exists on this frequency under SPWM operation.
To the best of the authors’ knowledge, there are few papers
discussing about the power quality improvement of the
dual-generator DC power system. However back-to-back
converter, which is with a rectifier and an inverter sharing a
common DC-link capacitor, is similar to our system. The power
sharing between the rectifier and the inverter is 1:1 (generation
and motoring).
In [19] and [20], Shen investigated reducing the ripple
current in a back-to-back converter by synchronizing the
carriers' phase based on band-pass filters and phase angle
measurement. However, the performance highly relies on the
quality of the band-pass filter (BPF), which limits the
application of this method. Compared to it, our paper proposed
In [21], Qin investigated reducing the ripple current in a
back-to-back converter. The method is also based on Fourier
analysis. However, the component in the frequency of fc is not
considered. As discussed in our paper, fc component will be
significant when the AC input voltages are high.
In this paper, we will fill the gap and develop a harmonic
cancellation scheme for multi-converter connected DC grid
with SVPWM implemented within AC/DC power converters.
Although it requires extensive math, the developed cancellation
scheme can be implemented in a convenient way. Furthermore,
although this paper focuses on dual-generator dc power system,
the research results of harmonic minimization scheme can also
be used for back-to-back converters [19][21]. The comparison
between this paper and previous work is given in TABLE I, and
the main contribution of this paper is as below:
• Propose a simplified mathematical model for both the
first and the second switching harmonics on the DC-link
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

capacitor . The calculation burden of microprocessor is idc

significantly released. The magnitude can be derived only
with the measurements of dq- voltages and currents, which is
ia
practical for the real-time controller.
• Based on the simplified models, a cancellation method of

DC bus
ib
DC-link harmonics was proposed. The algorithm shows that
variable phase-shift angle should be applied according to the ic
power range.
TABLE I PMSG
COMPARISON WITH PREVIOUS WORK
(a)
[16] [17] [18] [19] [21] This paper
Different f0 √ √ √ √ V2 β V3
SVPWM operation √ √ √ (010) (110)
Consider fc component √
Need no BPF √ √ √ √ √ 𝑽𝒓𝒆𝒇

V6 V1
The paper is organized as follows: Section II gives basic
(011) (100) α
mathematical analysis of switching harmonics based on Double
Fourier Method. Section III presents simplified mathematical
models for both the first and second switching harmonics on
capacitors to release the calculation burden of microprocessor. (001) (101)
V4 V5
Based on the simplified models, a new method is proposed to
suppress total harmonic in Section IV. With the proposed (b)
method, significant harmonic minimization can be achieved vc*
under any fundamental frequency and power range. Lastly,
Section V validates the proposed mathematical models and vb*
suppression method with both simulation and experimental
va*
results.

II.MATHEMATICAL MODEL FOR DC-BUS CURRENT HARMONIC idc

ON TWO-LEVEL CONVERTER UNDER SVPWM OPERATION
In order to develop a scheme to suppress dc voltage/current
harmonics within a multi-converter-fed DC grid, it is essential
to have detailed information of these harmonics either through
measurement or proper mathematic modelling. In this paper,
we assume that the power converters are standard two-level
converters as shown in Fig. 2a. Although measurement of
harmonics can be implemented using band-pass filters [16],
these filters are difficult to extract the required switching
frequency harmonics due to the existence of side-band
harmonics as shown in Fig. 3. In this section, relations of
DC-link harmonics and SVPWM modulation will be identified
and a mathematical model will be developed.
Fig. 2b shows an example of SVPWM operation when a
reference voltage (Vref) locates between two adjacent voltage
vectors V1 and V3. There, V1, V3 and two zero vectors V0, V7
synthesize the reference Vref. Fig. 2c illustrates how the
reference voltage is transformed into on/off signals for each of
the three inverter legs by comparing with a triangular carrier
wave. If the triangular carrier waveform is above the phase
voltage reference vx* (x=a,b,c), the corresponding leg will be
with ‘1’ state, i.e. the upper switch within this leg is “on”.
Otherwise, the corresponding leg will be with ‘0’ state, i.e. the
upper switch within this leg is “off”.
(000) (100) (110) (111) (110) (100) (000)
V0 V1 V3 V7 V3 V1 V0
(c)
Fig. 2 Two-level three-phase converter and SVPWM operation. a) Topology, b)
Switching vectors and voltage reference, c) Modulation waveform.
Fig. 3 shows the spectrum of idc which is the current flowing
into a dc-link capacitor from the converter side. fc is the
switching frequency. A few of side band components, fc±3f0 for
instance, can be identified. Here, f0 is the fundamental
frequency of the AC side voltage. Within the dual-channel
power generation system in Fig.1, the fundamental frequencies
f0 are normally of different values as generators are driven by
different shafts running at different speeds. Considering the fact
that a harmonics suppression scheme is essentially to use
harmonics from one generation system to compensate the other,
the targeted harmonics should be of the same frequency in two
different generation channels. The difference of f0 makes
suppression of side-band harmonics a very challenging task in
that sense. For that reason, this paper will focus on suppressing
components on fc and 2fc specifically.

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

Magnitude (A) 80 component of frequency fc+3f0. Using (1)–(4), the dc-bus

currents generated from one phase leg can be derived as
60 k 
idc ( t ) = iack  ( t ) sf k  ( t )
40 fc 2fc+6f0    4k   
 K 0,1 cos  4 f 0 t + 2  +  +  + cos  
20 2fc
   3    (5)
fc-3f0 fc+3f0 I ac  
=    cos  2 ( mf + ( n + 1) f ) t +   k   
0 2     
+   K m, n 
c 0 m , n
0 20 40 60 80 100 
+ cos  2 ( mf c + ( n − 1) f 0 ) t +  m , n   

Frequency (kHz) k
 m =1n =−    
Fig. 3 idc spectrum.
[𝑘] [𝑘]
where 𝜎𝑚,𝑛 and 𝜑𝑚,𝑛 (k=0,1,2) are given as
Neglecting the switching-frequency harmonics, AC currents  2k  
for a two-level converter can be written as  mk,n = m ck  + ( n + 1)   +  + n (6)
 3 
2k  
k 
( t ) = I ac cos  2 f 0t +  + 2k 
3  
iac (1)
 mk,n = m ck  + ( n − 1)   + + n
3 
(7)
where Iac is the amplitude of AC currents, f0 is the fundamental 
frequency, β is the phase angle (with reference to the electrical Thus, the DC-link current (idc) is essentially to add the
generator back-EMF in this paper), k=0, 1 and 2 is to represent contribution from the three-phase legs with (5). Components of
phase A, B and C respectively. Assuming the positive direction idc at the frequency of ifc+jf0 can be expressed as
2
I
idc ,i , j ( t ) = ac K i , j −1  cos  2 ( if c + jf 0 ) t +  i, j−1 
of current idc is from the converter to the dc-link capacitor as k
shown in Fig. 2a, the switching function for each phase leg can 2 k =0
 
be expressed with double Fourier expansion [23] as 2
(8)
I
+ ac K i , j +1  cos  2 ( if c + jf 0 ) t + i, j+1 
k
 2k 
sf   ( t ) = K 0,1 cos  2 f 0 t +  +  +  
k

3 
2 k =0
 where i and j mean orders of main and sideband harmonic of
 
 m 2 f t +   k 
 c c (

(2) ) DC currents of the converter respectively. The equation (8)
gives a detailed knowledge of each harmonic. For example,
+   K m, n cos  2k   
 + n  2 f 0 t +  +  +
substituting i=1 and j=3 into (8), harmonic on fc+3f0 can be
m =1n =− 
  3   expressed. (6) – (8) will be the basis for our harmonic
[𝑘]
cancellation scheme development.
where fc is the switching frequency, 𝜃𝑐 is phase angle of the
triangular carrier signal for each leg, α is the phase angle III. MATHEMATICAL ANALYSIS OF THE FIRST AND SECOND
between fundamental AC currents and voltages (i.e. power SWITCHING HARMONICS
factor angle), Km,n is the harmonic amplitude using the Bessel
function of the first kind. As discussed in [23], for As discussed in Section II, harmonics on the first and second
asymmetrical regular sampling SVPWM, the switching switching frequencies (fc and 2fc) have potential of cancellation.
function for a single phase leg can be expressed by In this section, some simplified models to calculate these two
     harmonic components are introduced.
 sin ( qm , n + n )    J n  3 qm , n M  + 2 cos  n  J n  3 qm , n M   
 2    4
6    6   4 
   A. Harmonic on fc
 
 1  qm , n   
 n   n    3   3  

 + sin   cos   sin   J0  qm , n M  − J 0  qm , n M    (3) Using (6) - (8), the first switching frequency harmonic can be
 n  2   6    4   4 
 2     n  0  given with i=1, j=0 as
 
  1        
sin ( qm , n + k )  cos ( n + k )  sin ( n + k ) 
2
K1,−1 cos  2 ( f c ) t + c  −  

I ac
idc,1,0 ( t ) =
    k
4   n+ k  2  2  6  
K m,n = 2 
+      
m  k =1   3      3   2 k =0
( k − n )   J k  qm , n M  + 2 cos  ( 2n + 3k )  J k  qm , n M    
(9)
    4   6  4    
   2
 
+ ac K1,1 cos  2 ( f c ) t + c  +  

  I k
 1       
 sin ( qm , n + k )  cos ( n − k )  sin ( n − k )  
  n−k

 2  2  6

  2  
+     k =0

 (kk=1n )   J  3 q M  + 2 cos  2n − 3k   J  3 q M   
  k
   4   ( )  
k  
 
Considering no phase shift of carrier signals among three
 
m,n m,n
   6  4 [1] [2] [3]
 
  phase legs, i.e., 𝜃𝑐 = 𝜃𝑐 = 𝜃𝑐 = 𝜃𝑐 , the equation (9) can
Where be rewritten as
f0 3I
qm, n = m + n
(4) idc ,1,0 ( t ) = ac K1, −1cos  2 ( f c ) t +  c −  
fc 2
(10)
3I ac
In (3) and (4), m and n mean the order of main and sideband + K1,1cos  2 ( f c ) t +  c +  
harmonic for a single phase leg respectively. For instance, 2
when m=1 and n=3, Km,n means the magnitude of harmonic

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

where the coefficients K1,-1 and K1,1 can be obtained according  3    3 

K1 = J1  M  + 2cos J1  M
to (3), which are
 4  6  4 

 
 
    3     3    1   
  
 6 sin ( q1, −1 − 1) 2   J −1  4 q1, −1 M  + 2 cos  6  J −1  4 q1, −1 M    sin ( k − 1)  
           6   k −1  6  (18)

  1        
sin ( q1, −1 + k )  cos ( k − 1)  sin ( k − 1) 
  

(11) +  
 k =3,5,7    3      3  
 
 k −1  2  2  6

   Jk  M  + 2cos ( 3k − 2 )  J k  M 
6   4  


4 +  
    4    
K1, −1 = 2  k =3,5,7...    3      3    
   Jk  q1, −1 M  + 2 cos  ( 3k − 2 )  J k  q1, −1 M   
   4   
   6  4     1   


 − sin ( k + 1)  
  1          
sin ( q1, −1 + k )  cos ( k + 1)  sin ( −1 − k )    6   k +1  6 

+



 −1 − k  2  2  6  

+ 
 k =1,3,5    3      3  

 k =1,3,5...   3      3     J M  + 2cos ( 3k + 2 )  J k  M 
   
  k  4 1, −1 
J q M + 2 cos ( 2 + 3k )  k 
J q1, −1 M 
     k  4   6   4  
 
   6  4    
  From (18), it can be noticed that K1 is dependent on M only.
  Using (17) and sinusoidal sum-to-product formula, the equation
      3      3   
 6 sin ( q1,1 + 1) 2   J1  4 q1,1 M  + 2 cos  6  J1  4 q1,1 M    (10) can be simplified as
           
  f  
  1

       
sin ( q1,1 + k )  cos ( k + 1)  sin ( k + 1)  
(12) idc ,1,0 ( t )  0 I ac K1 sin  cos  2 f c t +  c −  (19)
 
 k + 1  2   2   6   fc  2
4 +      
K1,1 = 2  k =1,3,5...    3      3 where, α is power factor angle, and is determined by several
   J q M + 2 cos  ( 2 + 3k )  J k  q1,1 M    
   k  4 1,1   6   4  
    facts such as machine parameters, output power and so on. The

  amplitude of this component is proportional to the term f0/fc,
          
 sin ( q1,1 + k )  cos (1 − k )  sin (1 − k ) 
1
   which means that in the case of high-speed operation of PMSG,
+

1 − k  2  2  6 
 k =  
  3       3    the magnitude of idc,1,0 is not neglectable. For a power generator
   J q M + 2 cos  ( 2 − 3k )  J k  q1,1 M   
3,5,7...

   k  4 1,1   6   4  
  
in Fig. 1, the term sinα can be obtained from measurements of

dq- currents and voltages of electrical generators using the
formula of vector product, which is
idq  vdq vq id − vd iq
The sine function terms in (11) and (12) with q1,n=1+nf0/fc sin  = = (20)
can be approximately expressed using Taylor series with first idq vdq id + iq2 vd2 + vq2
2

order.
From (20), it can be seen that the term sinα can be obtained
    f 
sin ( q1, n + k )  = sin (1 + k ) + n 0   without requirement of any machine parameters. Substituting
 2  2 fc 2  (20) to (19) gives
(13)
  f0  f vq id − vd iq  
 
 sin (1 + k )  + cos (1 + k )   n  idc ,1,0 ( t )  0 K1 cos  2 f c t +  c −  (21)
 2  2 fc 2 fc vd + vq
2 2
 2
where k is an odd number in (11) and (12). Considering n=1 and The equation (21) shows a simplified model of the first
-1, (13) becomes switching harmonic. It can be seen that the phase angle of this
    f  harmonic is only determined by phase angle of the carrier
sin ( q1, −1 + k )   −cos (1 + k )  0  (14) signal (θc). By shifting switching time of each converter, this
 2  2  fc 2
component can potentially be suppressed actively.
    f 
sin ( q1,1 + k )   cos (1 + k )  0  (15)
Furthermore, considering the fact that PMSG for aerospace
 2   2  fc 2 application is normally running at high speed and
Meanwhile, f0/fc approximately equals to 0, hence flux-weakening is thus required. Under these conditions, the
q1,1  q1, −1  1 (16) modulation index M is always saturated and remains constant.
Thus, K1 is a constant value.
Substituting (14) - (16) into (11) and (12) gives Fig. 4 shows the magnitude of this harmonic (Idc,1,0) using a
1 f0 permanent magnet generation system developed in [21]. As can
K1, −1  − K1, −1  K1 (17)
3 fc be seen in Fig. 4a, the magnitude of harmonic idc,1,0 decreases
Where while output power increases. A higher magnitude can also be
noticed with increased rotor speeds. This is because higher
rotor speed represents higher fundamental frequency f0. From
(21), it can be seen that the magnitude is proportional to
fundamental frequency f0.
Fig. 4b shows the comparison between detailed model (10)
and simplified models (21). It can be noticed that the

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

discrepancies are less than 0.1A. However, the simplified K 2  K 2,1  K 2, −1

model is much easier to implement in a controller.  
 
25     3      3   
20 k rpm  − 6  J1  2 M  + 2 cos  6  J1  2 M   
24 k rpm          
28 k rpm
20  
32 k rpm  1          
  sin ( 2 + k )  cos ( k + 1)  sin ( k + 1)    (24)
   k +1  2  2  6 
2 +   
Ihf12 (A)
(A)

15
 2  k =1    3      3    
   J M  + 2 cos  ( 2 + 3k )  J k  M  
Idc,1,0

   k  2   6   2  
   
10  
 
  1        
  1 − k
sin  ( 2 + k ) cos (1 − k ) sin (1 − k ) 
5  2  
 2   6   
+  
   
  3      3    
 M  + 2 cos  ( 2 − 3k )  J k 
k =1
0   J M  
0 10 20 30 40 50    k  2   6   2  
   
 
P (kW)
With (23) and sum-to-product formula of sinusoidal
(a)
functions, the magnitude of second carrier harmonic can be
22 expressed as
Accurate idc ,2,0 ( t )  3I ac cos  K 2 cos  2  ( 2 f c ) t + 2c  (25)
20 Simplified
For a two-level converter, the converter power P at the
AC-side terminal can be given as
(A)
hf12 (A)

18 I V 3I MVdc
20.5 P = 3 ac ac cos = ac cos (26)
IIdc,1,0

2 2 4
16 where Vac is the magnitude of converter phase voltage, M is the
20
modulation index of the converter. From (26), we have
14 4P 4I
19.5 I ac cos = = dc (27)
0 2 4 6 8 10 3MVdc 3M
12
0 10 20 30 40 50
P (kW) Substituting (27) to (25), it gives
4I K
(b) idc ,2,0 ( t )  dc 2 cos ( 4 f c t + 2 c ) (28)
Fig. 4 Magnitude of first switching harmonic. a) Magnitudes under different M
rotor speeds, b) Comparison between simplified model (21) and accurate model
(10) when rotor speed is 32 k rpm.
where Idc is the dc component of the current idc on dc-bus. From
(28), it is important to point out that the magnitude of the
B. Harmonic on 2fc second carrier harmonic current is proportional to dc-link
current (Idc) and its phase angle of this component is determined
The second switching frequency harmonic also plays a
by the angle of carrier signal (θc) only. This means that the
significant role in harmonic spectrum. A simplified model of
phase angle of second carrier harmonic on DC-bus current can
second switching harmonic will be introduced in this section.
be controlled by carrier signal θc within the AC/DC converter.
This harmonic component can be derived by substituting i=2,
If the phase angles are shifted by 180o between the two AC/DC
j=0 into (8) as
2 converters, the second carrier frequency harmonic will cancel
I
idc ,2,0 ( t ) = ac K 2, −1 cos  2  2 ( f c ) t + 2 c  −  
k
each other fully if they are of the same magnitudes.
2 k =0
  Comparison between magnitudes of simplified model and
(22)
I ac 2 original model is given in Fig 5. The modulation index is set as
+ K 2,1 cos 2  2 ( f c ) t + 2 c + 
 k  
2 k =0
  1.10 (0.95 × 2/√3) which reflects the fact that PMSG always
works under flux weakening operation. It can be seen that the
Considering no phase shift on carrier signals among three
[1] [2] [3] harmonic component (Idc,2,0) from detailed equation (23) and
legs, i.e., 𝜃𝑐 = 𝜃𝑐 = 𝜃𝑐 = 𝜃𝑐 , it can be rewritten as simplified one (28) matches very well, especially when the
3I
idc ,2,0 ( t ) = ac K 2, −1cos  2  2 ( f c ) t + 2 c −   power is higher than 10kW. With the power increases, the
2 power factor α will move towards to zero with more active
(23)
3I power from electrical generator while the reactive power for
+ ac K 2,1cos  2  2 ( f c ) t + 2 c +   de-fluxing kept almost the same. From (23), two components
2
on the right side of the equation are symmetric, this means
The coefficients K2,-1 and K2,1 can be derived with (3).
improvement of the accuracy of the simplification. In practice,
Using similar approximation technique as in the previous
we are more interested in high-power operation and this is the
section considering fc is much larger than f0, we have

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

operation region where power quality issues should be more IV. CAPACITOR HARMONIC MINIMIZATION METHOD
carefully addressed. Using models developed in section III, a harmonic
50 minimization scheme can be proposed. The aim of this scheme
is to achieve an optimal trade-off solution to suppress the first
20 k rpm
24 k rpm and second carrier harmonics at the same time. This is achieved
40 28 k rpm using an active phase shift concept. The system to be studied is
32 k rpm
Simplified
with two generators with two AC/DC converters as shown in
dc,2,0 (A)

30
(A)

Fig. 7.
IIhf12

15 PMSG
20 1
10 DC bus
idc{1}
10 5

Idc{1}* PWM1
0 Local
0 2 4 6 8 10 controller 1
0 icap
0 10 20 30 40 50
P (kW) I1{1}, I2{1} θc{1}

Fig. 5 Comparison of second carrier harmonic component from accurate (23) System Phase shift
and simplified (28) models with increased generator power and speed. controller controller
idc{2}
I1{2}, I2{2} θc{2}
C. Discussion DC-link
Idc{2}* PWM2 capacitor
From Fig 4 and 5, it can be seen that the first switching Local
controller 2
harmonic Idc,1,0 decreases when output power increases.
However, the second switching harmonic shows an opposite
phenomenon, i.e. it will increase with more power output. This
means that the first switching harmonic tends to be dominant PMSG + -
2 Vdc
under low power range while the second switching harmonic
tends to be dominant with an increased power. Furthermore, the Fig. 7 Control diagram of a dual-generator power system.
harmonic cancellation can be achieved by shifting the carrier
signal phase angle θc for different converters. However, the first To minimize the voltage ripple on DC capacitor, weighted
and second switching frequency harmonics require different total harmonic distortion of the capacitor should be analyzed,
phase shift angles. Thus, an optimized trade-off scheme needs which is
to be identified, which will be discussed in the next section. 2
 Ik  
From (21) and (28), the switching frequency harmonics can
be derived and the implementation of such equations in a
 I hf = f c
 
k =1  f k 
(29)
generation system is shown in Fig. 6, where I1 and I2 represent
where Ihf is total harmonics on dc capacitor, fk is frequency of
Idc,1,0 and Idc,2,0 respectively.
kth order harmonic. Ik is the magnitudes of summed kth order
harmonic from all converters connected to the common dc bus.
Coupling to
HP or LP shaft As discussed in Section II, only harmonics on fc and 2fc have
PMSG To DC-link
capacitors
potential of cancelling due to no impacts from the fundamental
ω r θr idc frequencies. Hence, these two components are extracted from
{

abc Vdc
dq
SVPWM
(29)
2 2
𝑣𝑑2 + 𝑣𝑞2 vd* vq* θr  I fc   I 2 fc  I 2 fc 2
I hf _1,2 = f c   +  = I fc +
2
(30)
-
Vdc/2
× - id* θc
 fc   2 fc  4
M PI
× Decoupled Here, we use subscripts to represent the frequencies of the
d- and q- harmonic for convenience. Using phase angles of the
2 − 𝑖𝑞2
𝑖𝑚𝑎𝑥 axis current components given from (21) and (28), (30) can be rewritten as
- control
iq* iq*
iref1 PI
-
Low-pass ωr
Idc filter

Harmonic
Equ (28) Equ (21) calculation

I2 I1

Fig. 6 Control diagram and harmonic calculation scheme of single PMSG

system.

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

2 to (33), optimized Δθc is also determined by magnitudes of the

 {1}   {2}  {2}  
1 4 I    c − 2  + I1    c − 2 
{1}
1
first and second harmonics from each converter.
I hf _1,2 = To give a clear result of how Δθc varies, it is assumed that
2
( ) ( )
2 two converters supply the same power to the downstream loads.
+ I 2{1} 2 c{1} + I 2{2} 2 c{2} Fig. 9 shows the optimized phase-shift angle versus output
power of single converter using (33). It can be seen that when
( ) + 4 ( I ) + 8I I
2 2
{1}
1 4 I1
{2}
1
{1} {2}
1 1 cos  c output power is lower than a barrier (around 10kW in this case),
= (31) the optimized Δθc is 180°constantly. As discussed, under low
( ) + ( I ) + 2I I
2 2
2 + I {1} {2} {1} {2}
cos 2 c
2 2 2 2 power region, the first switching component dominates the
harmonic spectrum. Thus using (21), phase difference of the
1 4 I 2 I 2 cos  c + 8I1 I1 cos  c
{1} {2} 2 {1} {2}
first switching harmonics from two converter is θc{2}- θc{1}.
=
( ) ( ) ( ) ( )
2 2 2 2
2 +4 I1{1} + 4 I1{2} + I 2{1} + I 2{2} − 2 I 2{1} I 2{2} Hence, a 180-degree phase difference between carrier signals
of two converters can reduce the most of first switching
Here, θc{1} and θc{2} are phase angles of carrier signals. I1{1}, harmonic harmonics.
I2 , I1{2}, and I2{2} are the magnitudes of harmonics from each
{1}
With higher output power, the second switching components
converter. Here, superscripts in square brackets ‘{}’ are become dominant. Then the phase-shift angle needs to be
indexes of the converters. Δθc is the phase-shift angle between adjusted. With higher output power, optimized Δθc keeps
two converters which is dropping down. According to (28), phase difference between
 c =  c{2} −  c{1} (32) the second switching harmonics of two converters is
(2θc{2}-2θc{1}). When output power is high enough to be
In (31), variable fc, I1{1}, I2{1}, I1{2} and I2{2} are determined by dominant, it is the equation (2θc{2}-2θc{1}) = 180°to be used in
power flow supervised by system level controller. Minimizing the system controller. Then, phase difference between the two
the total switching frequency harmonics Ihf_1,2 can thus be carrier signals (θc{2}- θc{1} or Δθc) should be 90°. This explains
achieved by adjusting Δθc. Using basic law of quadratic and why the optimized angle drops approximately to 90°.
triangle function, the optimized phase-shift angle can be
200
derived from (31) as
 −1  I1{1} I1{2}  I1{1} I1{2}
cos  − {1} {2}  , 1 160 I1{1} I1{2}
=1
Δθc (°)

{1} {2}
  I2 I2  I2 I2 I 2{1} I 2{2}
Δ c =  (33)
 I1{1} I1{2} 120
 , 1
 I 2{1} I 2{2}
Based on (33), a phase-shift controller can be designed as 80
0 10 20 30 40 50
P (kW)
shown in Fig. 8. Meanwhile, the optimized Ihf_1,2 can also be
derived based on (31).
Fig. 9 Optimized Δθc when power sharing ratio between two converters is 1:1.
Local controller 1
Coupling to
HP or LP shaft
PMSG
θc{1} θc{1}
ω r θr idc
50
{

abc Vdc

I1{1}, I2{1}
dq
SVPWM
vd* vq* θr

+
𝑣𝑑2 + 𝑣𝑞2

0°
θc{2}
Vdc/2 -
× - id* θc
M PI
90°
Equ. Δθc
× Decoupled
d- and q-
40 180°
hf12 (A)

+
2 − 𝑖𝑞2
𝑖𝑚𝑎𝑥 axis current
-
iref1
-
PI
iq* iq* control

ωr
(33) Opitimzed
(A)

Low-pass
IIhf_1,2

Idc filter

Equ (28) Equ (21)

Harmonic
calculation 30
I2 I1

From Fig.6
Coupling to
HP or LP shaft
PMSG 20
ω r θr idc
{

abc Vdc
dq
SVPWM
𝑣𝑑2 + 𝑣𝑞2 vd* vq* θr
Vdc/2
× - id*
-
θc
Equ. Ihf_1,2 10
M
×
PI
Decoupled
d- and q- 0 10 20 30 40 50
iref1 PI
iq*
2
𝑖𝑚𝑎𝑥 − 𝑖𝑞2

iq*
-
axis current
control (31) P (kW)
-
ωr
Idc
Low-pass
filter Fig. 10 Ihf_1,2 versus output power of single converter.
Equ (28) Equ (21)
Harmonic
calculation I1{2}, I2{2}
I2 I1

From Fig.6 The active phase shifting mechanism developed with (33)
Local controller 2 ensures that the Ihf_1,2 has been suppressed in its lowest level.
Fig. 8 Phase-shift controller. Fig. 10 shows a comparison of Ihf_1,2 with different phase-shift
strategies. It can be seen that when no phase shift applied, i.e.,
In the implementation, the phase angle of carrier signal of Δθc =0, Ihf_1,2 shows highest value across the entire power
one converter can be fixed (say converter 1), and the carrier range. The case with phase shift of 90°works better than that of
signal for the other converter is adjusted using (33). According 180°under high power region, while the case with phase shift
angle of 180° works better under low power region. The

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

proposed mechanism combines benefits from the both angles TABLE II

SIMULATION PARAMETERS
and achieves the lowest Ihf_1,2. Meanwhile, the proposed method
only adjusting phase angle of the carrier signal, and the Category Parameters Values
phase-shift time is neglectable compared to fundamental PMSG PMSG parameters From [22]
signals. Therefore, system performance (i.e., efficiency and Rotor speeds (n{1} and n{2}) 20kRPM & 30kRPM
stability) will not be affected. Switching frequency (fc) 32kHz
Maximum modulation index (M) 1.04 (0.9 × 2/√3)
dq-current controller bandwidth 1000 Hz
V. SIMULATION AND EXPERIMENT RESULTS DC-link DC-link capacitance 400µF
Load power 20kW to 80kW
A. Simulation results
Simulation is implemented with Matlab/Simulink and Fig. 11a-d show the spectrums of dc bus current from
PLECS to validate the performance of the proposed harmonic converter 1 and 2 (idc{1} and idc{2}), capacitor currents (icap) under
models and cancellation method. In the simulation, the PMSGs different power output and phase-shift strategies. Converter 1
are the one developed in [22], which is a 6-pole PMSG. Some and 2 generate the power with a sharing ratio of 1:1.
basic control parameters are given in the Table II. Switching Components on fc and 2fc are highlighted in red.
frequencies of the two converters are the same, which is 32 Fig. 11a shows the result when power generated from single
kHz. This is designed based on carrier-fundamental frequency converter is 5kW and there is no phase-shift angle between two
ratios of 20:1 at the maximum speed (32 k rpm, 1600Hz). carrier signals. Switching harmonics on both buses just sum up

(A)
(A)

50 50

i{1}
i{1}

dc
dc

0 0
0 20 40 60 80 100 0 20 40 60 80 100
(A)
(A)

50 50
i{2}
dc
i{2}
dc

0 0
0 20 40 60 80 100 0 20 40 60 80 100

24.92A 10.41A
50 15.39A
icap(A)

icap(A)

50 15.38A

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (kHz) Frequency (kHz)
(a) (b)
(A)
(A)

50 50
i{1}
i{1}

dc
dc

0 0
0 20 40 60 80 100 0 20 40 60 80 100
(A)
(A)

50 50
i{2}
i{2}
dc

dc

0 0
0 20 40 60 80 100 0 20 40 60 80 100

17.95A 74.62A 4.43A

(A)

50
icap(A)

50 14.40A
cap
i

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (kHz) Frequency (kHz)
(c) (d)
Fig. 11 Current spectrums of idc{1}, idc{2}, and icap. a) 5kW, no phase shift, b) 5kW, optimized phase shift, c) 40kW, no phase shift, d) 40kW, optimized phase shift.

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

on capacitor, hence significant components show on both the (a)

frequencies of fc and 2fc. For idc{1} and idc{2}, magnitudes of the 12

Reduction of Ihf (%)

first switching harmonics are higher than the second ones. In
Fig. 11b, to mostly supress first switching harmonic, a phase 8
difference of 180°is applied according to Fig 9. Compared to
Fig. 11a, the first switching harmonic on capacitor is reduced
4
from 24.92A to 10.41A (58% reduction). However, the second
switching harmonic are with no difference because phase angle
of the second harmonic is 2θc. 180°phase difference on carrier 0
20 40 60 80
signals results in 360° difference on the second harmonic,
P (kW)
which is same as with no phase shift. (b)
In Fig. 11c and d, output power per converter is increased to Fig. 12 Ihf variation after applying the proposed method.
40kW. To clearly show the magnitude changes in fc and 2fc, the
DC components from each converter (idc{1} and idc{2}), which are Compared to the control with no optimization, the proposed
148.15A, are not fully shown in these two figures. As analysed method needs some extra computations, i.e., fc and 2fc harmonic
in former sections, the second switching harmonic become estimation and phase-shift angle calculation. The total
more significant compared to the first one. Therefore, a larger computation time of these actions are 13.56μs. In the
second order component shows in Fig. 11c because there is no experiment, they are implemented in every 10ms (dc-link
phase shift action here. To maximum the suppression on both current control loop). Therefore, the computation intensity of
the first and the second switching harmonics, optimized phase the proposed method is not a problem for the system.
shift angle is set as 92.2°according to (33). The results are idc1
shown in Fig. 11d. The second switching harmonic on capacitor icap
is significantly reduced from 74.62A to 4.43A (96% reduction).
Such an almost 100% suppression is because proportional Transfomer
relationship between the second switching harmonic and output Grid & Variac idc2
power, which has been proved in (28). When the two sources Programmable
generate same power, magnitudes of the second switching AC source
harmonics from two converters are almost same. With a nearly
180°(2×90°) phase difference between two components, the (a)
suppression is quite effective as shown here. Converter 1 Capacitor Converter 2 Transformer
Apart from the second switching harmonics, the first
switching harmonic is also suppressed as shown in Fig. 11d. It
is reduced from 17.95A to 14.40A (19.8% reduction). Although
90°phase difference is not as good as 180°, it still helps weaken
the sum component of this order.
It is interesting to notice that the phase shift will not create
unappreciated increase on the icap components other than in fc
and 2fc (i.e., fc±3f0{1} and fc±3f0{2}). This can be observed in Fig.
11a-f. In simulation, we calculated Ihf using (33) under different
output power and the power sharing ratio is 1:1, as shown in
Fig. 12a. Ihf is suppressed on the whole range. To clearly show Resistor load Variac Programmable
AC source
the reduction of Ihf, Fig. 12b shows the reduction ratio after
(b)
applying the proposed method. The reduction is around 3%
Fig. 13 Experiment setup. (a) Schematic drawing. (b) Physical
under low power range (i.e. 20kW). When the output power is
drawing.
higher, (i.e., 80kW), the reduction is more significant which
can be 8%. Therefore, the capacitance can be chosen based on
TABLE III. Experiment parameters
the output power range. If the output power is higher, a lower
capacitance can be chosen. Parameters Values
110 DC-link capacitance 4.4 mF
Line inductance 1 mH
100 DC-bus voltage 270V
Ihf (A)

Output power 400W-4kW

90 Frequencies of AC side 50 Hz and 70 Hz
Without suppression Voltages of AC side 250 V
80 With suppression Switching frequency 2 kHz
70 Maximum modulation index 1.04 (0.9 × 2/√3)
20 40 60 80 Dead-time 500 ns
P (kW)

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

4
Fig. 14 show the experiment results of a single converter Simplifed model (fc)
when f0=70Hz. AC side current (ia), dc-bus current before

Magnitude of harmonics (A)

Experiment results (fc)
capacitor (idc) and spectrum of it are given in Fig. 14a and b. 3 Simplifed model (2fc)
Experiment results(2fc)
Output power of them are 400W and 2kW, respectively. When
P=400W, harmonic on fc is higher than that on 2fc. However,
with output power increased to 2kW, harmonic on 2fc becomes 2
more dominant on the spectrum.
Derived from increasing power step by step, magnitudes of
1
harmonics on both fc and 2fc are listed in Fig. 14c. Meanwhile,
simplified models of proposed in Section III are also drawn in
it. Compare the values in experiment results and simplified
0
models, errors between them are always less than 0.1A, which 0 500 1000 1500 2000 2500
validate the effectiveness of the proposed mathematical P (W)
models. (c)
Fig. 15a and b show the currents on dc bus of two converters Fig. 14 Experiment on single converter when f0=70Hz. (a)
(idc{1} and idc{2}) and capacitor (icap) when output power of the P=400W. (b) P=2kW. (c) Comparison between experiment and
dual-generator system is 400W. Power for each converter are mathematical model.
the same, 200W. In Fig. 15a, no phased shift is applied.
Harmonics in fc and 2fc are 0.93A and 0.62A. Here, first order (50A/div)
idc{1}
harmonic is not neglectable compared to second order
harmonic. According to (33), 180-degree phase shift is applied idc{2}
to mainly supress the first switching harmonic. The result is icap
shown in Fig. 15b. Harmonic on fc is suppressed to 0.12A, Vdc (100V/div)
which has a suppression of 87%. Meanwhile, harmonic on 2fc is
almost same.
fc: 0.93A 2fc: 0.62A Spectrum of idc
ia (2A/div)
(25A/div) fc-3f0 fc+3f0

idc
(a)
(25A/div)
{1} (50A/div)
idc
fc: 0.563A Spectrum of idc
2fc: 0.524A (1A/div) idc{2}
fc-3f0 fc+3f0
icap
Vdc 180° (100V/div)

(a)
fc: 0.12A 2fc: 0.62A Spectrum of idc
ia
(2A/div)
(25A/div) fc-3f0 fc+3f0
idc
(25A/div) (b)
Fig. 15 Currents on DC buses and capacitor when output power is 400W. (a) No
2fc: 2.732A Spectrum of idc phase shift. (b) With optimized phase-shift angle.
fc: 0.421A
(1A/div)
fc-3f0 fc+3f0 When the output power of the dual-generator system
increases to 4kW, the second order harmonic become more
dominant in capacitor current spectrum which is shown in
(b) Fig. 16a. Where harmonics on fc and 2fc are 0.53A and 4.63A.
With an optimized phase shift angle which is 92.6°according to
(33), the experiment result is shown in
Fig. 16b. There, the second order harmonic is supressed to
0.22A (95.2% suppression). Meanwhile, first order harmonic is
also reduced from 0.53A to 0.38A (28.3% suppression). Both
the first and second switching harmonics are supressed, which
is as same as the simulation result shown in Fig. 11.

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

(50A/div) [4] K. Kwon, J. Lee and S. Min, "Motor and Transmission

idc{1} Multiobjective Optimum Design for Tracked Hybrid Electric
idc{2} Vehicles Considering Equivalent Inertia of Track System," in
icap IEEE Transactions on Transportation Electrification, vol. 7, no.
4, pp. 3110-3123, Dec. 2021.
Vdc (100V/div) [5] K. Ni, C. Gan, Y. Hu, D. T. Lagos, R. Qu and N. D.
Hatziargyriou, "Emulated Stator Voltage-Oriented Vector
Control of DFIM-SPS With Coupling Effect Elimination for
fc: 0.53A 2fc: 4.63A Spectrum of idc Electric Ship Applications," in IEEE Transactions on
(2A/div)
fc-3f0 fc+3f0 Transportation Electrification, vol. 7, no. 3, pp. 1615-1627, Sept.
2021.
[6] J. Westman, R. Hadidi, J. C. Fox and J. Leonard, "Modeling of a
(a) Reciprocating Engine-Generator Set for
idc {1} (50A/div) Controller-Hardware-in-the-Loop Testing of Island Microgrid
Control Systems," in IEEE Transactions on Industry
idc{2} Applications, vol. 57, no. 4, pp. 3377-3384, July-Aug. 2021.
icap [7] H. B. Enalou, X. Lang, M. Rashed and S. Bozhko, "Time-Scaled
92.6° (100V/div)
Emulation of Electric Power Transfer in the More Electric
Vdc
Engine," in IEEE Transactions on Transportation Electrification,
Spectrum of idc vol. 6, no. 4, pp. 1679-1694, Dec. 2020, doi:
fc: 0.38A 2fc: 0.22A (2A/div) 10.1109/TTE.2020.2999400.
[8] H. Wang and F. Blaabjerg, "Reliability of Capacitors for
fc-3f0 fc+3f0 DC-Link Applications in Power Electronic Converters—An
Overview," in IEEE Transactions on Industry Applications, vol.
50, no. 5, pp. 3569-3578, Sept.-Oct. 2014, doi:
(b)
10.1109/TIA.2014.2308357.
Fig. 16 Currents on DC buses and capacitor when output power is 4kW. (a) No [9] F. D. Kieferndorf, M. Forster and T. A. Lipo, "Reduction of
phase-shift. (b) With optimized phase-shift angle. DC-bus capacitor ripple current with PAM/PWM converter," in
IEEE Transactions on Industry Applications, vol. 40, no. 2, pp.
607-614, March-April 2004, doi: 10.1109/TIA.2004.824495.
VI. CONCLUSION
[10] K. Choi, S. Kim, S. Jung and R. Kim, "Generalized Switching
In this paper, simplified mathematical models for dc-link Modification Method Using Carrier Shift for DC-link Capacitor
harmonics on the AC-DC converter are investigated. The RMS Current Reduction in Real-Time Implementation," in IEEE
calculation burden of microprocessor is significantly released. Transactions on Industrial Electronics, vol. 66, no. 8, pp.
For the first switching harmonic, magnitude can be achieved 5992-6001, Aug. 2019, doi: 10.1109/TIE.2018.2873160.
with the measurements of dq- voltages and currents. For the [11] X. Lu and F. Z. Peng, "Theoretical analysis of DC link capacitor
current ripple reduction in the HEV DC-DC converter and
second switching harmonic, the magnitude is proportional to
inverter system using a carrier modulation method," 2012 IEEE
the DC component of the output current. Energy Conversion Congress and Exposition (ECCE), Raleigh,
Based on the simplified models, a cancellation method of NC, 2012, pp. 2833-2839, doi: 10.1109/ECCE.2012.6342376.
dc-link harmonics was proposed. Under low power range, [12] M. Mellincovsky, V. Yuhimenko, M. M. Peretz and A.
phase-shift angle is set as 180°to suppress the first switching Kuperman, "Analysis and Control of Direct Voltage Regulated
harmonic mostly. When the output power increases, the Active DC-Link Capacitance Reduction Circuit," in IEEE
phase-shift angle is adjusted from 180° to 90° actively to Transactions on Power Electronics, vol. 33, no. 7, pp. 6318-6332,
suppress both the first and second switching harmonics. July 2018, doi: 10.1109/TPEL.2017.2746079.
Finally, simulation and experiment were implemented to [13] W. Liu, K. Wang, H. S. Chung and S. T. Chuang, "Modeling and
Design of Series Voltage Compensator for Reduction of DC-Link
verify the validity of the proposed harmonic models and the
Capacitance in Grid-Tie Solar Inverter," in IEEE Transactions on
cancellation method. Power Electronics, vol. 30, no. 5, pp. 2534-2548, May 2015, doi:
10.1109/TPEL.2014.2336856.
REFERENCES [14] L. Zhang and X. Ruan, “Control schemes for reducing the second
[1] Abdollahi and G. B. Gharehpetian, "A 20-Pulse Autotransformer harmonic current in two-stage single-phase converter: An
Rectifier Unit for More Electric Aircrafts," in IEEE Journal of overview from dc-bus port-impedance characteristic”, in IEEE
Emerging and Selected Topics in Power Electronics, vol. 9, no. 3, Transactions on Power Electronics, vol. 34, no. 10, pp. 10341–
pp. 2992-2999, June 2021. 10358, Oct. 2019.
[2] A. A. Recalde et al., "Energy Storage System Selection for [15] L. Zhang, X. Ruan, and X. Ren, “One-cycle control for
Optimal Fuel Consumption of Aircraft Hybrid Electric Taxiing electrolytic capacitor-less second harmonic current compensator”,
Systems," in IEEE Transactions on Transportation in IEEE Transactions on Power Electronics, vol. 33, no. 2, pp.
Electrification, vol. 7, no. 3, pp. 1870-1887, Sept. 2021. 1724–1739, Feb. 2018
[3] X. Lang, T. Yang, C. Li, H. B. Enalou, S. Bozhko and P. Wheeler, [16] G. Gohil et al., "Modified Discontinuous PWM for Size
"A Dual-Channel-Enhanced Power Generation Architecture Reduction of the Circulating Current Filter in Parallel Interleaved
With Back-to-Back Converter for MEA Application," in IEEE Converters," in IEEE Transactions on Power Electronics, vol. 30,
Transactions on Industry Applications, vol. 56, no. 3, pp. no. 7, pp. 3457-3470, July 2015, doi:
3006-3019, May-June 2020. 10.1109/TPEL.2014.2339392.

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.
 This article has been accepted for publication in IEEE Journal of Emerging and Selected Topics in Power Electronics. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/JESTPE.2022.3188912

IEEE Journal of Emerging and Selected Topics in Power Electronics

[17] D. Zhang, F. Wang, R. Burgos, R. Lai and D. Boroyevich,
"DC-Link Ripple Current Reduction for Paralleled Three-Phase
Voltage-Source Converters With Interleaving," in IEEE
Transactions on Power Electronics, vol. 26, no. 6, pp.
1741-1753, June 2011, doi: 10.1109/TPEL.2010.2082002.
[18] C. Wang, T. Yang, P. Kulsangcharoen and S. Bozhko, "An
Enhanced Second Carrier Harmonic Cancellation Technique for
Dual-Channel Enhanced Power Generation Centre Applications
in More-Electric Aircraft," in IEEE Transactions on Industrial
Electronics, vol. 68, no. 7, pp. 5683-5692, July 2021, doi:
10.1109/TIE.2020.2998755.
[19] L. Shen, S. Bozhko, G. Asher, C. Patel and P. Wheeler, "Active
DC-Link Capacitor Harmonic Current Reduction in Two-Level
Back-to-Back Converter," in IEEE Transactions on Power
Electronics, vol. 31, no. 10, pp. 6947-6954, Oct. 2016, doi:
10.1109/TPEL.2015.2511304.
[20] L. Shen, S. Bozhko, C. I. Hill, and P. Wheeler, “DC-Link
Capacitor Second Carrier Band Switching Harmonic Current
Reduction in Two-Level Back-to-Back Converters,” in IEEE
Transactions on Power Electronics, vol. 33, no. 4, pp. 3567–
3574, 2018.
[21] Z. Qin, H. Wang, F. Blaabjerg and P. C. Loh, "Investigation into
the control methods to reduce the DC-link capacitor ripple
current in a back-to-back converter," 2014 IEEE Energy
Conversion Congress and Exposition (ECCE), Pittsburgh, PA,
2014, pp. 203-210, doi: 10.1109/ECCE.2014.6953395.
[22] S. Bozhko et al., "Development of Aircraft Electric Starter–
Generator System Based on Active Rectification Technology,"
in IEEE Transactions on Transportation Electrification, vol. 4,
no. 4, pp. 985-996, Dec. 2018, doi: 10.1109/TTE.2018.2863031.
[23] D. G. Holmes and T. Lipo, Pulsewidth Modulation for Power
Converters. 568 New York, NY, USA: Wiley, 2003.

© 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 03,2022 at 13:32:58 UTC from IEEE Xplore. Restrictions apply.