energies

Article
Partial Stray Inductance Modeling and Measuring of
Asymmetrical Parallel Branches on the Bus-Bar of
Electric Vehicles
Chengfei Geng ID
, Fengyou He *, Jingwei Zhang and Hongsheng Hu
School of Electrical and Power Engineering, China University of Mining and Technology, Xuzhou 221116, China;
ge-cumt-001@163.com (C.G.); hangjingwei@cumt.edu.cn (J.Z.); hhssgd@126.com (H.H.)
* Correspondence: hfy_cumt@263.net; Tel.: +86-516-8013-9898

Academic Editors: Joeri Van Mierlo and Omar Hegazy

Received: 16 August 2017; Accepted: 26 September 2017; Published: 1 October 2017

Abstract: In order to increase the power rating of electric vehicles, insulated gate bipolar translator
(IGBT) modules with multiple power terminals are usually adopted. The transient current sharing of
the same polarity power terminals is related to the stray inductance in the branches of the bus-bar.
Based on the laminated bus-bar of a three-phase inverter in the electric vehicles that consists of
asymmetrical parallel branches, this paper investigates the transient current imbalance sharing
caused by the asymmetrical stray inductance in the parallel branches of the bus-bar from the view of
energy storing and releasing of stray inductance for the first time. Besides, the partial self-inductance
and mutual-inductance model of the parallel branches is set up. Finally, a high-precision partial stray
inductance measurement method is proposed, and the accuracy of the partial stray inductance model
for asymmetrical parallel branches is verified by experimental tests.

Keywords: laminated bus-bar; electric vehicles; partial self-inductance; partial mutual inductance;
asymmetrical branches

1. Introduction
In recent years, electrical vehicles (EVs) have gained unprecedented evolution in consideration
of the current energy crisis and environmental protection [1]. For an EV inverter, high efficiency,
high power density, low harmonics, small volume, and low cost are essential [2–4]. Due to the
low stray inductance, the laminated bus-bar has now been widely used in power electronics and
converter products [5–7]. With power devices continuing to move upwards in current levels and
switching frequency, laminated bus-bars have been attracting increasing interest from both industry
and academia for the system benefits they exhibit [8]. The over-voltage spikes across the semiconductor
during switching directly results from the commutation loop stray inductance. The high di/dt values
produced during commutations by these new fast switching devices and diodes are currently the main
source of conducted and radiated noise emissions, which usually interferes with nearby equipment
through conduction or radiation [9,10]. For an EV inverter, the position of the small signal control
circuit is often very close to high voltage, large current components such as the bus-bar or the power
devices considering the limited space in an enclosure. The strength of Electro Magnetic Interference
(EMI) emission is associated with transient current distribution on the bus-bar. So, for reliable operation
of the control circuit in an EV inverter, the analysis, extraction, and evaluation of the stray inductance
for a laminated bus-bar are very important.
Due to the increasing power of EV inverters, IGBT modules with multiple power terminals
are often adopted. Moreover, the transient current sharing performance between power terminals
of the same polarity is related to the stray inductance of the branches. In prior research, the most

Energies 2017, 10, 1519; doi:10.3390/en10101519 www.mdpi.com/journal/energies

Energies 2017, 10, 1519 2 of 16

focus has been on optimizing the bus-bar structure based on the circuit topology [6,7,11]. Partial
inductance theory is applied to build the model of bus-bar stray inductance [12–14]. The Partial
Element Equivalent Circuit (PEEC) method [15,16] is adopted to construct the model and investigate
the transient current distribution in the laminated bus-bar. Furthermore, with consideration of the
current path, a partial self-inductance and mutual-inductance model of a H-bridge bus-bar has been
built [17]. All the modeling and analysis are based on a single power electronics topology. But, the
transient imbalance current sharing caused by the asymmetrical stray inductance in bus-bar parallel
branches is comparatively scarce. This paper deeply investigates the transient imbalance current
sharing arising from the asymmetrical stray inductance in bus-bars from the view of stray energy
storing and releasing for the first time. The partial self-inductance and mutual-inductance model of
the asymmetrical parallel branches is also set up to facilitate transient current analysis.
In order to evaluate the stray inductance of the bus-bar and verify the modeling accuracy, many
measuring methods have been proposed by scholars and engineers. These include gap voltage
measurement based on the IGBT turn-on process [18], time domain reflectometry (TDR) method [19],
resonant capacitance measurement methods [6], stray inductance integral arithmetic extraction method
for bus-bar based on turn-on/off transient process [20], network analyzer measurement [17], etc.
However, all these methods mentioned above focus on the total stray inductance of the current
communication loops rather than the partial inductance. This paper proposes a high precision
method to measure the partial inductance in the bus-bar, and the measuring accuracy is also analyzed
considering the spatial transient electromagnetic field. Experimental evaluation of partial stray
inductance extraction including direct and indirect measurement verifies the proposed model and
measuring method.

2. Inverter Layout and Equivalent Circuit of the Bus-Bar

2.1. Inverter Layout Consideration

The main drivers for the development of EVs can be summarized to be cost, power density,
specific volume, efficiency, and reliability. In this design, the power rating of the EV inverter is 200 kW.
The inverter is assembled using individual building blocks serving the system, including switch
(power device IGBT), energy storage (DC link capacitor), current distributor (copper bus-bar structure),
and cooling system (liquid cooled heat sink). In addition, sensor functions (Hall elements) and control
circuits are integrated. This setup is connected using screws and is supported by mechanical parts
and protected by an enclosure, which is shown in Figure 1a. The inner structure of the inverter is
shown in Figure 1b, in which the 1.2-KV/1400-A half bridge IGBT module is used. A low-inductance,
high-frequency film capacitor is used to replace the conventional electrolytic bulk capacitors to reduce
size and enhance the reliability.
High-speed switching devices produce lots of EMI noise during the fast on-off process, which
usually interferes with nearby equipment through conduction or radiation [8–10]. Typical problems
caused are measurement errors in sensors, false control signals, and interference in the gate voltages of
switching devices [8]. Excessive noise can deteriorate control, reduce efficiency, and cause malfunctions
of system. In this design, the control circuit is on the topside of the bus-bar. High impedance circuits
are sensitive to EMI, such as the signal conditioning circuit of analog to digital (AD) acquisition.
The source of EMI is associated with the transient current changes in the bus-bar. Simulation plays an
important role in understanding the operating principle of the whole system and the functioning of
some key components because it is more instructive and can save time and cost. Previous simulation
platforms only concern analysis of the control algorithm. Parasitic elements are usually omitted to
save simulation time [21]. So, in order to predict and evaluate the EMI/Electro Magnetic Compatibility
(EMC) performance accurately and to further guide the layout of some key components in a limited
space for EV inverters, a precise simulation platform that includes the main parasitic components
should be built up.
Energies 2017, 10, 1519 3 of 17
Energies 2017, 10, 1519 3 of 16
Energies 2017, 10, 1519 3 of 17

Current
DC capacitor sensor
DSP Control Current
DC capacitor Board sensor
DC-
DSP Control C
Board B DC-
C
A
B
DC+ A
DC+
Water cooled
output
enclosure Water cooled
output
enclosure Water cooled input
Water cooled input
(a)
(a)
DC capacitor
DC capacitor
Lamiated bus bar
Lamiated bus bar
DC-
N DC-
N
Gate Driver
P C Gate Driver
P C
B
B
b
b
A
DC+ A
DC+

IGBT
IGBTmodules
modules

(b)
(b)
Figure
Figure
Figure 1.
1. 1.The
Thestructure
The structureofofthe
structure EV
EV motor
thethe
of motor
EV system
driverdriver
motor system (MDS).(MDS).
(MDS).
system (a)External
(a) External structure;
structure;
(a) External (b)(b) Internal
Internal
structure; (b)
structure.
structure.
Internal structure.

2.2.2.2.
2.2.
EquivalentCircuit
Equivalent
Equivalent
CircuitofofBus-bar
Circuit of
Bus-barConsidering
Considering Stray
Bus-bar Considering
Stray Inductance
Inductance
Stray Inductance
TheThestructure
structureandandcircuit
circuitprinciple
principle ofof the half bridge IGBT module adopted in this design is is
The
shown
structure
in Figure
and
2.
circuit
There
principle
are 5 power
the the
halfhalf
ofterminals on
bridge
bridge
the
IGBT
IGBT,
IGBT
modulemodule
terminals 9
adopted
adopted
and
thisin this
in represent
11
design design
is shown
collector
shown
in Figure in 2.
Figure
There 2.are
There are 5 terminals
5 power power terminals on theterminals
on the IGBT, IGBT, terminals
9 and 119 and 11 represent
represent collectorcollector
of the
of the upper
of the upper IGBT,
IGBT, terminals
terminals 10 and
1012and 12 represent
12 represent emitter
emitter of the
oflower lower
the lower IGBT
IGBT and terminal
and terminal 8 represents
8 represents
upper IGBT, terminals
the middle terminal. 10 and represent emitter of the IGBT and terminal 8 represents the
the middle
middle terminal.
terminal.
9/11
5 9/11
5
12
12 11 4
11 10 6 4
10 9 3
9 8 5 6
3 8
8 5 4 NTC 8
6 3
4 NTC
67 3
7 7 1
1
2 7 1
1 2
2
2
10/12
10/12
Figure 2. The structure and circuit principle of the IGBT module.
Figure 2. The
Figure 2. The structure
structure and
and circuit
circuit principle
principle of
of the
the IGBT
IGBT module.
module.
Energies 2017, 10, 1519 4 of 16
Energies 2017, 10, 1519 4 of 17

detailed structure
The detailed structureof ofthe
thelaminated
laminatedbus-bar
bus-barisisshown
shownininFigure
Figure3. 3.
TheThe power
power terminal
terminal P and
P and N
N are
are connected
connected to anode
to the the anode and cathode
and cathode of theofDCthe
busDC bus capacitor.
capacitor. A, B, and A,CB,are
and
theCoutput
are the output
terminals
terminals
of of the
the inverter. Theinverter. The power
power terminals terminals
9 and 9 and
11 on the 11 onbus
positive thearepositive
connectedbus to
arethe
connected
Collector to
of the
Collector
upper of the
IGBT. Theupper
power IGBT. The power
terminals 10 andterminals
12 on the10negative
and 12 onbusthe
arenegative
connectedbus to
arethe
connected to the
Emitter of
Emitter
lower of the lower IGBT.
IGBT.
This paper focuses on the stray inductance of parallel branches of the bus-bar, without the
consideration of the straystray inductance
inductance in in the
the capacitor
capacitor and
and IGBT
IGBT module.
module. The simplified stray
inductance model of the three phase inverter is shown shown in in Figure
Figure 4a.
4a. The detailed
detailed stray
stray inductance
inductance
model with the consideration of the asymmetrical parallel branch of every phase on the bus-bar is
Figure 4b.
shown in Figure 4b. Lσ1σ1 and
and LLσ2σ2represent
representstray
stray inductance
inductance from
from positive
positive DCDC busbus terminal
terminal P to
and 11
terminals 9 and 11 respectively.
respectively. LLσ3σ3 and Lσ4σ4 represent
represent the
the stray
stray inductance
inductance from negative DC bus
terminal N to terminals 12 and and 10 respectively.
10 respectively.

P
N
12

11

10

Structure of the bus-bar.

Figure 3. Structure bus-bar.

Lσ1
11
LσB1
Lσ2
LσA1 LσC1 c
P 9

SA1 SB1 SC1 SA1

Udc Udc
8

SA2 SB2 SC2 SA2

N
Lσ3 12
LσA2 LσB2 LσC2 e
Lσ4 10

(a) (b)
Figure 4. Equivalent circuit of every phase leg. (a) Simplified circuit; (b) Detailed circuit.
Figure 4. Equivalent circuit of every phase leg. (a) Simplified circuit; (b) Detailed circuit.

3. Transient Analysis and Modeling of Asymmetrical Branches

3. Transient Analysis and Modeling of Asymmetrical Branches
3.1. Transient
3.1. Analysis
Transient Analysis
There are
There are eight
eight switch
switchstates
statesfor
foraathree-phase
three-phaseinverter
inverterand
andonly
only one
one power
power device
device is permitted
is permitted to
to change state at any time under normal operation conditions. The double
change state at any time under normal operation conditions. The double pulse test is used pulse test is used to test
to test the
the IGBT
IGBT over-voltage
over-voltage and stray
and stray inductance
inductance of theofbus-bar,
the bus-bar, meanwhile,
meanwhile, every shares
every phase phase shares
the same thetesting
same
testing principle, and this paper defines phase B as the example. Figure 5 is the basic
principle, and this paper defines phase B as the example. Figure 5 is the basic principle of the test principle of the
test circuit. The i 1, i2, i3, and i4 represent the current in the branches of Lσ1, Lσ2, Lσ3, and Lσ4. Furthermore,
circuit. The i1 , i2 , i3 , and i4 represent the current in the branches of Lσ1 , Lσ2 , Lσ3 , and Lσ4 . Furthermore,
taking Proximity
taking ProximityEffect Effect into
into consideration,
consideration, coupled
coupled inductance
inductance exists among
exists among the fourthe fourbranches.
current current
branches.
 Energies 2017, 10, 1519 5 of 17
Energies 2017, 10, 1519 5 of 16
Energies 2017, 10, 1519 5 of 17

Lσ1 i1
Lσ1 i1
i2 Mσ1_2 Lσ2
P i2 Mσ1_2 Lσ2
P
Mσ1_3 S1
Mσ1_3 S1
Udc
Udc Mσ1_4 Mσ2_3
Mσ2_3 S2
Mσ1_4 Load
S2
Load
Mσ2_4
Mσ2_4
Lσ3 i3
N Lσ3 i3
N Mσ3_4 iload
Lσ4 Mσ3_4 i4 iload
Lσ4 i4

Figure 5. The test circuit of every phase leg.

Figure5.5.The
Figure The test
test circuit ofevery
circuit of everyphase
phaseleg.
leg.

The transient analysis is based on the two branches of Lσ1 and Lσ2, and the waveforms of the
The transient analysis
is isbased
basedon on thetwotwo branches
branches of of LLof
σ1 and Lσ2,σ2and thethe waveforms of the
doubletransient
The analysis
pulses test is depicted in Figurethe6. The time intervals –t1, tL3–t
σ1t0and , and waveforms
4, and t7–t8 represent steady-
of the
double
double pulses
pulses test
test isis depicted
depicted in
in Figure
Figure 6. The
6. time
The intervals
time of
intervals t 0–t1, t3–t4, and t7–t8 represent steady-
of t –t , t –t , and t –t represent
state of conducting and blocking. This paper focuses on the transient process of turn-on and turn-off
0 1 3 4 7 8
state of conducting
steady-state and blocking.
of conducting andintervalThis paper
blocking. focuses on the transient process of turn-on and turn-off and
which refers to the time of This
t1–t3 paper
and t4–t focuses
7, and on thethe transient
current in process of turn-on
t8–t9 shares the same
which refers to the time interval of t1–t3 and t4–t7, and the current in t8–t9 shares the same
turn-off which refers
characteristics withto the time
current interval
in the of tinterval.
t1–t3 time 1 –t3 and t4 –t7 , and the current in t8 –t9 shares the same
characteristics with current in the t1–t3 time interval.
characteristics with current in the t1 –t3 time interval.

VVgg t t

½·
½·itotal
itotal
ii11
t t
i2i2

tt00 t1t1t2tt23t3 t4t45t65t67 t7 t8 t8t9 t9

Figure 6. The waveforms of the double pulse test.

Figure
Figure6.6.The
Thewaveforms
waveformsofofthe
thedouble
doublepulse
pulsetest.
test.
3.1.1. Turn-Off Transient
3.1.1. Turn-Off Transient
3.1.1. Transient
At tAt
1 , S1 receives the turn-off signal and and thebus-bar
bus-bar can be equivalent theto the circuit depicted
At tt11,, S1
S1 receives
receives thethe turn-off
turn-offsignal signal andthe the bus-barcancanbebeequivalent
equivalentto to circuit
the depicted
circuit in in
depicted
in Figure
Figure 7,7,ininwhich
which itotal
i represents the turn-off current. According to Kirchhoff’s Current Law, the
total represents the turn-off current. According to Kirchhoff’s Current Law, the
Figure 7, in which itotal represents the turn-off current. According to Kirchhoff’s Current Law, the
branch current
branch current satisfies i1 +i1i+2 i=2 =i3i3++i4i4 == iitotal
satisfies , meanwhile, the voltage in stray inductance of each branch
branch current satisfies i1 + i2 = i3 + i4 = total itotal, ,meanwhile,
meanwhile,the
thevoltage
voltageinin
stray inductance
stray inductance of of
each branch
each branch
satisfies u
satisfies u=
L1uL1 =L2
satisfies L1 u
= u and
L2 and
uL2 and uu u L3 =
= u
u
L3L3 = uL4
L4 .
L4.
.
The turn-off process of an IGBT can be regarded as the stray inductance energy releasing process,
so the following analysis is based on magnetic field energy stored in the stray inductance. At t1 , the
energy stored in branches Lσ1 and Lσ2 are given as
(
WLσ1 (t1 ) = 12 Lσ1 i12 (t1 )
(1)
WLσ2 (t1 ) = 21 Lσ2 i22 (t1 )

WLσ1 (t1 ) and WLσ2 (t1 ) are determined by the current and inductance in the two branches at t1 ,
meanwhile, S1 starts to turn off and itotal starts to decrease. Hence, the energy stored in Lσ1 and
Energies 2017, 10, 1519 6 of 16

Lσ2 begins to release, moreover, the speed of releasing is determined by the stray inductance in the
two branches and the speed of turn-off. If the energy stored in Lσ1 and Lσ2 is released completely
at t3 simultaneously, that means WLσ1 (t3 ) = 0 and WLσ2 (t3 ) = 0, and the current direction in the two
branches depicted in Figure 7 does not change from t1 to t3 . If the energy stored in Lσ1 and Lσ2 does
not release completely at t3 simultaneously, that means WLσ1 (t2 ) 6= 0 or WLσ2 (t2 ) 6= 0 and one of the
two branches must be released completely prior to the other. This study supposes the energy stored in
Lσ2 releases completely at t2 , namely i2 (t2 ) = 0. From t1 to t2 , the current i1 and i2 decrease gradually
shown in Figure 8 (t1 –t2 ) but the current direction does not change. At t2 , i1 (t2 ) = itotal (t2 ) and Lσ1 can
be equivalent to a power source, and the direction of potential is shown in Figure 7, so the branch
of Lσ2 provides the path of energy release for Lσ1 that means the current i2 changes direction from t2
and increases at the same slope as i1 shown in Figure 8 (t2 –t3 ), but they share the contrary direction.
At t3 , the current i1 and i2 are equal in magnitude and opposite in direction and itotal = 0, moreover,
neglecting the influence of tail current, the IGBT is closed completely. Due to the asymmetrical stray
inductance in branches Lσ1 and Lσ2 , a loop current is caused after turning off the IGBT.
At the turn-off transient, due to uL1 = uL2 , the variation ratio of i1 and i2 can be calculated by

Energies 2017, 10, 1519 di1 L 6 of 17

= σ2 (2)
di2 Lσ1

Lσ1 uL1 i1

Mσ1_2
uL2
Lσ2
i2
Mσ1_3
itotal
Mσ2_3
Mσ1_4
Mσ2_4
i3 uL3 Lσ3
Mσ3_4
uL2
Lσ4 i4

Energies 2017, 10, 1519 Figure 7. The equivalent circuit of the turn-off transient. 7 of 17
Figure 7. The equivalent circuit of the turn-off transient.

The turn-off process of ani1 IGBT can be regarded i1 as the stray inductance energy releasing
process, so the following analysis is based on magnetic field energy stored in the stray inductance.
itotal itotal
At t1, the energy stored in branches Lσ1 and Lσ2 are given as
i2
 1 i22
WL 1 (t1 ) 2 L 1i1 (t1 )
(t1-t2)  1 (t2) (1)
WL (t1 ) L 2i22 (t1 )
i1  2
2 i1
WLσ1(t1) and WLσ2(t1) are determined iby the current and inductance
total itotal in the two branches at t1,
meanwhile, S1 starts to turn off and itotal starts to decrease. Hence, the energy stored in Lσ1 and Lσ2
begins to release, moreover, the ispeed 2 i2
of releasing is determined by the stray inductance in the two
branches and the speed of turn-off. If the energy stored in Lσ1 and Lσ2 is released completely at t3
simultaneously, that means W(t (t-t
Lσ12 3)3=) 0 and WLσ2(t3) = 0, and (t3the
) current direction in the two branches
depicted in Figure 7 does not change from t1 to t3. If the energy stored in Lσ1 and Lσ2 does not release
completely at t3 Figure 8. The current
simultaneously, thatdirection
meansof Wparallel branches at the turn-off transient.
Lσ1(t2) ≠ 0 or WLσ2(t2) ≠ 0 and one of the two branches
Figure 8. The current direction of parallel branches at the turn-off transient.
must be released completely prior to the other. This study supposes the energy stored in Lσ2 releases
completely
3.1.2. Turn-Onat t2Transient
, namely i2(t2) = 0. From t1 to t2, the current i1 and i2 decrease gradually shown in Figure
8 (t1–t2) but the current direction does not change. At t2, i1(t2) = itotal(t2) and Lσ1 can be equivalent to a
At t4, S1 receives the open signal, due to the reverse recovery characteristic of the freewheeling
power source, and the direction of potential is shown in Figure 7, so the branch of Lσ2 provides the
diode inside S2, and S2 is equivalent to a short-circuit. The equivalent circuit is depicted in Figure 9.
path of energy release for Lσ1 that means the current i2 changes direction from t2 and increases at the
 i2 i2
(t2-t3) (t3)
Energies 2017, 10, 1519 7 of 16
Figure 8. The current direction of parallel branches at the turn-off transient.

3.1.2.
3.1.2. Turn-On
Turn-On Transient
Transient
AtAtt4 ,t4S1
, S1receives
receivesthe
theopen
opensignal,
signal, due
due to
to the
the reverse
reverse recovery
recoverycharacteristic
characteristicofof
the freewheeling
the freewheeling
diode inside S2, and S2 is equivalent to a short-circuit. The equivalent circuit is depicted
diode inside S2, and S2 is equivalent to a short-circuit. The equivalent circuit is depicted inin
Figure 9. 9.
Figure

Lσ1 uL1
i1
Mσ1_2
uL2
i2 Lσ2

Mσ1_3
itotal
Mσ2_3
Mσ1_4
Mσ2_4
uL3
i3 Lσ3
Mσ3_4
uL4 i4
Lσ4

Figure 9. The equivalent circuit of the turn-on transient.

Figure 9. The equivalent circuit of the turn-on transient.

i1 (ti14(t)4and
) andi2i(t
2(t44)) are
are mainly associatedwith
mainly associated withthe
thecommutation
commutation between
between Lσ1 Landσ1 and Lσ2 during
Lσ2 during t3–t4. At t3 –t4 .
Attt44, ,ififthe
theenergy
energystored
storedininLσ1 Lσ1and andLσ2Lis
σ2 released
is releasedcompletely, that
completely, means
that W
means W
Lσ1 (t 4) =
Lσ1 4 0
(t and
) = 0Wand
Lσ2 (tW4) =
Lσ2 0,(t4 )
also i (t ) = 0 and i (t ) = 0. Starting from t , i and i increase rapidly from
= 0, also i1 (t4 ) = 0 and i2 (t4 ) = 0. Starting from t4 , i1 and i2 increase rapidly from zero to the peak
1 4 2 4 4 1 2 zero to the peak of reverse
of recovery
reverse recovery current, the rising the
current, slope is related
rising slopetoisthe strayto
related inductance
the stray in the branch,inand
inductance thethe magnitude
branch, and the
and direction of current is shown in Figure 9. If the energy stored
magnitude and direction of current is shown in Figure 9. If the energy stored in Lσ1 and Lσ2 does not in L σ1 and L σ2 does not release
completely duringduring t3–t4, that
t3 –tmeans some loop current exists and i1(t4) = − i2(t4), as depicted in Figure
release completely 4 , that means some loop current exists and i1 (t4 ) = − i2 (t4 ), as depicted
10 (t 4). Starting from t4, due to the external voltage, i1 increases rapidly and i2 decreases to zero quickly,
in Figure 10 (t4 ). Starting from t4 , due to the external voltage, i1 increases rapidly and i2 decreases to
as depicted in Figure 10 (t4–t5), until i1(t5) = itotal(t5), as depicted in Figure 10 (t5). Then, starting from t5,
zero quickly, as depicted in Figure 10 (t4 –t5 ), until i1 (t5 ) = itotal (t5 ), as depicted in Figure 10 (t5 ). Then,
i1 and i2 keep the same direction and rise to maximum value at the same time. During the period of
starting from t5 , i1 and i2 keep the same direction and rise to maximum value at the same time. During
t6–t7, the reverse recovery current is at the vanishing stage and is equivalent to the load current iload.
the period of t6 –t7 , the reverse recovery current is at the vanishing stage and is equivalent to the load
Similar to the turning-off process, i1 and i2 may change direction, but the differential of the two
current iload . Similar to the turning-off process, i1 and i2 may change direction, but the differential of
currents
Energies 2017, can10,
also
1519be described by Formula (2). 8 of 17
the two currents can also be described by Formula (2).

i1 i1
itotal itotal
i2 i2
(t4) (t4-t5)
i1 i1 i1
itotal itotal itotal
i2 i2 i2
(t5) (t5-t6) (t6-t7)

Figure 10. The current direction of parallel branches at the turn-on transient.
Figure 10. The current direction of parallel branches at the turn-on transient.
3.2. Model of Stray Inductance
3.2. Model of Stray Inductance
Based on the above analysis, although the direction of current changes at the turn-on transient
Based on the above analysis, although the direction of current changes at the turn-on transient
or the turn-off transient, the position of the bus-bar does not change, and the dotted terminals of
or the turn-off transient, the position of the bus-bar does not change, and the dotted terminals of the
the branch equivalent inductance remains the same. Ultimately, the over-voltage caused by mutual
branch equivalent inductance remains the same. Ultimately, the over-voltage caused by mutual
inductance
inductancealso stays
also the
stays same.
the same.
The partial stray inductance model for Lσ1, Lσ2, Lσ3, and Lσ4 is as follows:

 u L1   Ls 1 M1 2 M1 3 M1 4   i1 
u L 2  M 2 1 Ls 2 M 23 M 2 4  d i2 
u    M M32 Ls 3
 
M 3 4  dt i3  (3)
 L3   3 1   
u M M M L i
Energies 2017, 10, 1519 8 of 16

The partial stray inductance model for Lσ1 , Lσ2 , Lσ3 , and Lσ4 is as follows:
     
u L1 Lsσ1 M12 M13 M14 i1
 u L2   M21 Lsσ2 M23 M24  d  i2 
= · · (3)
     
u L3 M31 M32 Lsσ3 M34  dt  i3
 
   
u L4 M41 M42 M43 Lsσ4 i4

The uL1 , uL2 , uL3 , and uL4 represent induced voltage of stray inductance Lσ1 , Lσ2 , Lσ3 , and Lσ4 ,
respectively, at the switching transient. The Lsσ1 , Lsσ2 , Lsσ3 , and Lsσ4 represent self-inductance in
branches Lσ1 , Lσ2 , Lσ3 , and Lσ4 . Meanwhile the M12 (M21 ), M13 (M31 ), M14 (M41 ), M23 (M32 ), M24 (M42 ),
and M34 (M43 ) represent the mutual-inductance for each branch.
While measuring the bus-bar stray inductance, only the transient voltage and current of the
bus-bar are available, so the measured values are all of equivalent stray inductance, and the
mathematical model is as follows.
     
u L1 Lσ1 0 0 0 i1
 d  i2 
 u   0 Lσ2 0 0   
 L2  
= · · (4)
 u L3   0 0 Lσ3 0  dt  i3 
 

u L4 0 0 0 Lσ4 i4

According to Equation (3) and (4), the bus-bar equivalent stray inductance is expressed as:
 h i h iT
di2 di3 di4


 L σ1 = Lsσ1 M12 M13 M14 · 1 di1 di1 di1

 h i h iT
· di di3 di4

 Lσ2 = M21 1
Lsσ2 M23 M24 1

di di2 di2
h i h 2 iT (5)


 L σ3 = M31 M32 Lsσ3 M34 · di
di
1 di2
di3 1 di4
di3
i h 3


 h iT
· di di2 di3

 L = M 1
σ4 41 M42 M43 Lsσ4 di di4 di4 1
4

For Equation (5), the differential is difficult to obtain from measurement at the turn-off transient.
Because of the excellent linearity at the turn-on transient, the differential can be replaced by difference.
 h iT iT
∆i2 ∆i3 ∆i4
h
di2 di3 di4


 1 di1 di1 di1 = 1 ∆i1 ∆i1 ∆i1

 iT iT
∆i1 ∆i3 ∆i4
h h
di1 di3 di4

1 = 1

∆i2 ∆i2 ∆i2

di di2 di2
h 2 iT iT (6)
∆i1 ∆i2 ∆i4
h
di 1 di2 di4



 di3 di3 1 di3 = ∆i3 ∆i3 1 ∆i3
 iT iT
∆i1 ∆i2 ∆i3
 h h
 di1 di2 di3

di4 di4 di4 1 = ∆i4 ∆i4 ∆i4 1

Based on Figures 7 and 9, the simplified equivalent circuit is shown in Figure 11.
The induced voltage in the switching transient of the stray inductance LσPc and LσNe meets:
" # " # " #
u Pc LsσPc MPN d itotal
= · (7)
u Ne MPN LsσNe dt itotal

where, uPc = uL1 = uL2 , uNe = uL3 = uL4 . LsσPc and LsσNe represent the self inductance and mutual
inductance of LσPc and (between) LσNe . The current matrices meet:
 
" # " # i1
itotal 1 1 0 0  i2 
= · (8)
 
itotal 0 0 1 1 i3

 
i4
 Energies 2017, 10, 1519 9 of 16
Energies 2017, 10, 1519 9 of 17

Based on Equations (7) and

 (8), the simplified
T inductance matrix
T meets as:
 di di di   i i3 i4 
 1 2 3 4   1 2 
  di#1 di1 "di1   i1 #i1 i1 
  −1
1 0
"  T T
LsσPc MPNdi1 = di3 di41 1   0 i1 0 i3 −1i4  1 0
 
 1    1 ·L · (9)
 0 0 1i 1 i i 
 
MPN L 0 1

sσNe
 di2 di2 di2   2 2 2

 T T 0 1 (6)

 di1 di2 di4   i1 i2 i4 
  di di 1    1 
where:   3 3
di3   i3 i3 i3
 di1 di2 Ldisσ1
TM M M T
 12 i1 13 i3 
i2 14
 3
 M 1 Lsσ2   M23 M24 1
L4 =di4 di21
 di
 M31 M32 Lsσ3 M34 4   i4 i4 i4  (10)

M41 M42equivalent
Based on Figure 7 and Figure 9, the simplified M43 Lsσ4circuit is shown in Figure 11.

LσPc uPc LσPc uPc

MPN itotal MPN itotal

Energies 2017, 10, 1519 10 of 17

LσNe uNe LσNe uNe

 Ls1 M 12 M 13 M 14 
M Ls 2 M 23 M 24 
L   21 
(a)  M 31 M 32 Ls 3 M 34  (b) (10)
 
Figure
Figure11.
11.The
Thesimplified
simplifiedequivalent  Min41inthe
equivalentcircuit
circuit Mswitching
the42switching s 4 
M 43 transient.
Ltransient.
(a)(a)
Turn
Turnoff
offtransient;
transient;(b)
(b)turn
turn
onontransient.
transient.
4. Measuring Method
The induced forinPartial
voltage Stray Inductance
the switching transient of the stray inductance LσPc and LσNe meets:
4. Measuring Method for Partial Stray Inductance
4.1. Basic Principle uPc   LsPc M PN  d itotal
4.1. Basic Principle u Ne   M PN LsNe   dt itotal (7)
The measuring principle for partial stray inductance is based on the inductance voltage integral
The measuring principle for partial stray inductance is based on the inductance voltage integral
method
where, uPc[20].
= uL1Under
= uL2, uhigh
Ne = ufrequency,
L3 = uL4. LsσPcthe
and stray
LsσNeinductance is determined
represent the by the
self inductance inductance
and outside the
mutual inductance
method [20]. Under high frequency, the stray inductance is determined by the inductance outside the
ofconductor, tending to
LσPc and (between) be. The
LσNe constant.
current The voltagemeet:
matrices and current waveform of the stray inductance at the
conductor, tending to be constant. The voltage and current waveform of the stray inductance at the
turn-off process of an IGBT is shown in Figure 12.
turn-off process of an IGBT is shown in Figure 12.  i1 
ito ta l 1 1 0 0 i2 
ito ta l  0 0 1 1  i3  (8)
 
i t  i4 
Based on Equations (7) and (8), the simplified inductance matrix meets as:

u t 
1
 1 0 
  
A  LsPc M PN B    1 1 0 0  1 1 0 
  
M Lt sNe   0 0 1 1
L
1 
i  0 (9)
 PN  
u t
 
t1

0 1 t
2

where:

Figure 12. The current, voltage waveform of inductance at the turn-off transient.
Figure 12. The current, voltage waveform of inductance at the turn-off transient.

Furthermore, the voltage on stray inductance depends on the changing rate of current, expressed
by following equation:

u t   L 
di
(11)
dt
Energies 2017, 10, 1519 10 of 16

Furthermore, the voltage on stray inductance depends on the changing rate of current, expressed
by following equation:
di
u(t) = Lσ · (11)
dt
By the calculation of the definite integral for Equation (11), the bus-bar partial stray inductance is
obtained and t1 –t2 represents the time of the turn-on transient or turn-off transient.
R t2
t1 u(t)dt
Lσ = (12)
i ( t2 ) − i ( t1 )

4.2. Measurement Error Analysis

Based on Equation (12), the measured value of partial stray inductance is determined by the
integral value of voltage during the period of t1 –t2 and the values of i(t2 ) and i(t1 ). So, the precise
partial stray inductance can be obtained once the integral value of the voltage and the two transient
current values are available.
The measurement diagram of stray inductance is shown in Figure 13. Because the bus-bar partial
stray inductance is very small, the voltage on the stray inductance at the switching transient is also very
small, so a passive probe is needed to raise the bandwidth and precision of the voltage measurement.
The probe of the oscilloscope is a high impedance circuit, vulnerable to the interference of an external
magnetic field, and the interference source is mainly the voltage on the measuring lead-wire caused
Energies 2017, 10, 1519
by
11 of 17
the spatial transient magnetic field.

Busbar
Measuring Oscilloscope
wire

i cl2
A c13
Lσ
Oscilloscope
cl1 probe

Figure 13.
Figure The measuring
13. The measuring method
method of
of partial
partial stray
stray inductance.
inductance.

The
The measuring
measuring wire
wire isis divided
divided into
into ccl1l1,, ccl2l2,, and
and ccl3l3, ,shown
shownin
inFigure
Figure 13.
13. According
According to
to the
the theory
theory
of partial inductance, the induced voltage on the measuring lead-wire caused by
of partial inductance, the induced voltage on the measuring lead-wire caused by the spatial transientthe spatial transient
magnetic
magnetic field
field can
can be
be expressed by:
expressed by:

(  ·Adl d+l   AA d· ldl

dA +  A  dlA) · dl)
R R R
d(
Verr =Verr 
cl1 cl 1 ccl2l 2 cl 3 cl3
(13)
(13)
dt
dt
represents the magnetic vector potential caused by the bus-bar
A represents bus-bar current
current and the
the magnitude
magnitude and
and
direction is decided by the Biot–Savart law [14]

µ JJ
vv RR ·dvdv
Z
AA= (14)
(14)
4
4π

J represents
representsthe thecurrent
currentdensity
densityvector.
vector.Based
Basedon on
Formula (14),(14),
Formula A shares the same
A shares direction
the same with
direction
the
withcurrent and isand
the current inversely proportional
is inversely to thetodistance
proportional fromfrom
the distance the bus-bar
the bus-barin magnitude. So, the
in magnitude. So,
measuring
the measuringprecision can be
precision raised,
can guaranteeing
be raised, the lead-wire
guaranteeing cl1 andccl1l2 are
the lead-wire andperpendicular to the bus-
cl2 are perpendicular to
bar, and cl3 is very far from the bus-bar. If the distance between cl3 and bus-bar exceeds a certain value,
the measuring voltage remains constant, which stands for the measured value with high precision.

5. Simulation and Experimental Results

Energies 2017, 10, 1519 11 of 16

the bus-bar, and cl3 is very far from the bus-bar. If the distance between cl3 and bus-bar exceeds a
certain value, the measuring voltage remains constant, which stands for the measured value with
high precision.

5. Simulation and Experimental Results

5.1. Parameter Extraction

Ansys Q3D software was used to extract the partial stray inductance of the laminated bus-bar
model in this study to verify the measured values of partial stray inductance. For extraction of the
partial stray inductance, a current source and sink must be set up. For a certain conductor, only one
current sink is allowed, but multiple current sources are permitted in the Q3D software. Taking the
proximity effect of the bus-bar into consideration, excitation is demanded for both a positive bus-bar
and a negative bus-bar to simulate the model. The positive bus-bar terminals 9 and 11 are set as
Source2 and Source1 and the capacitor anode is set as Sink1. Similarly, the negative bus-bar terminals
10 and 12 are set as Source4 and Source3 and the capacitor cathode is set as Sink2. The excitation source
settings and simulated current density of the bus-bar is shown in Figure 14. Because the excitation has
the opposite direction of the actual current, the simulated mutual inductance in the branches of the
positive bus-bar
Energies 2017, and negative bus-bar must be negative. The self inductance and mutual inductance
10, 1519 12 of of
17
the asymmetrical parallel branches are described in Table 1.

Sink1 Sink2

Source3
Source1

Source4
Source2

Figure 14. The excitation source settings and simulated current density of the bus-bar.
Figure 14. The excitation source settings and simulated current density of the bus-bar.
Table 1. Simulated partial stray inductance matrix.
Table 1. Simulated partial stray inductance matrix.
nH Lσ 1 Lσ 2 Lσ 3 Lσ 4
nH Lσ1 Lσ2 Lσ3 Lσ4
Lσ1 22.72 23.95 −1.56 −7.56
Lσ2 Lσ1 22.72
23.95 23.95
37.42 −1.56
−2.18 −7.56 −14.5
Lσ3 Lσ2 23.95
− 1.56 37.42
−2.18 −2.18
17.18 −14.5 17.48
Lσ4 Lσ3 − 7.56
−1.56 −14.5
−2.18 17.48
17.18 17.48 29.37
Lσ4 −7.56 −14.5 17.48 29.37
In order to verify the accuracy of the simplified model, the stray inductance LσPc and LσNe should
In orderbased
be extracted to verify
on the accuracy The
simulation. of the simplified
capacitor model,
anode theas
is set stray inductance
Source1, and theLσPcpositive
and LσNebus-bar
should
be extracted based on simulation. The capacitor anode is set as Source1, and the positive
terminals 9 and 11 are set as Sink1. Similarly, the negative bus-bar terminals 10 and 12 are set as bus-bar
terminalsand
Source2 9 and
the 11 are set cathode
capacitor as Sink1.isSimilarly, the The
set as Sink2. negative bus-bar
simplified terminals
model 10 and
excitation 12 are
source set as
settings
Source2 and the capacitor cathode is set as Sink2. The simplified model excitation source
and current density of the bus-bar are shown in Figure 15. The simulated self inductance and mutual settings and
current density of the bus-bar are shown in Figure 15. The simulated
inductance of the simplified branch shown in Figure 10 are described in Table 2. self inductance and mutual
inductance of the simplified branch shown in Figure 10 are described in Table 2.

source1 sink2

source2
sink1
 In order to verify the accuracy of the simplified model, the stray inductance LσPc and LσNe should
be extracted based on simulation. The capacitor anode is set as Source1, and the positive bus-bar
terminals 9 and 11 are set as Sink1. Similarly, the negative bus-bar terminals 10 and 12 are set as
Source2 and the capacitor cathode is set as Sink2. The simplified model excitation source settings and
current density of the bus-bar are shown in Figure 15. The simulated self inductance and mutual
Energies 2017, 10, 1519 12 of 16
inductance of the simplified branch shown in Figure 10 are described in Table 2.

source1 sink2

source2
sink1

source2
sink1

Figure 15. The simplified model excitation source settings and current density of the bus-bar.
Figure 15. The simplified model excitation source settings and current density of the bus-bar.
Table 2. Simulated stray inductance matrix.
Table 2. Simulated stray inductance matrix.

nH nH LσPc
LσPc LσNe
LσNe
LσPcLσPc 17.81
17.81 −1.44
−1.44
LσNeLσNe − 1.44
−1.44 16.46
16.46

The simplified inductance matrix calculated based on Equation (9) is shown in Table 3. Calculation
results are compared with the simulated results in Table 2. They have only 1 nH difference. That means
the calculated values can match the simulated values very well, and the simplified method is accurate.

Table 3. Calculated stray inductance matrix.

nH LσPc LσNe
LσPc 19.14 −1.51
LσNe −1.51 17.15

5.2. Experimental Measurement

In order to measure the branch stray inductance, the inverter testing platform is constructed as
shown in Figure 16. The measuring equipment and power device are shown in Table 4. It must be
noted that the oscilloscope must have the function of multi-channel digital integral that can measure
the stray inductance of the two parallel branches under the same conditions. The experimental test
is classified as the direct measuring method and the indirect measuring method. For the indirect
measuring method, a Rogowski coil is adopted to measure the transient current variation ratio in
the four branches. Meanwhile, using the simulation value of the stray inductance and Equation (6),
the equivalent stray inductance can be obtained. The direct measuring method can be realized by the
method proposed by the paper.

Table 4. The experimental equipment.

Equipment Parameter
IGBT module FF1400R12IP4
IGBT driver 2SP0320V2A0
Rogowski coil CWT30B
Voltage probe Agilent-10073C
Oscilloscope Agilent-MSO8064A
Inductive load About 7uH
the stray inductance of the two parallel branches under the same conditions. The experimental test is
classified as the direct measuring method and the indirect measuring method. For the indirect
measuring method, a Rogowski coil is adopted to measure the transient current variation ratio in the
four branches. Meanwhile, using the simulation value of the stray inductance and Equation (6), the
equivalent stray inductance can be obtained. The direct measuring method can be realized by
Energies 2017, 10, 1519
the
13 of 16
method proposed by the paper.

Laminated
Busbar capacitor
busbar

IGBT module Measuring

And driver wire

Rogowski coil

Figure
Figure 16.
16. The
The experimental
experimental platform.
platform.

5.3. Indirect Measurement Table 4. The experimental equipment.

Due to the excellent linearity Equipment

of the transient currentParameter
of the IGBT module at the turn-on transient,
the indirect measurement takes IGBT
themodule
turn-on transient asFF1400R12IP4
an example. In order to verify the consistency
Energies 2017, 10, 1519 14 of 17
of the transient current’s change IGBT
ratedriver 2SP0320V2A0
under different testing conditions, a double pulse test is applied
on S1 and S2. At first, according Rogowski coil leg testing circuit
to the phase CWT30B shown in Figure 5, S1 is tested with a
double pulse, then, upon switching the position of the load, S2 is tested also with the double pulse.
double pulse, then, upon switching Voltage the
probeposition of theAgilent-10073C
load, S2 is tested also with the double pulse.
The transient current i1, i2, i3, and i4 shown in Figure 17a,b, corresponding to the second turn-on of
The transient current i1 , i2 , i3Oscilloscope
, and i4 shown in Figure Agilent-MSO8064A
17a,b, corresponding to the second turn-on
lower module S2 and upper module S1, and their main difference is the current magnitude and
of lower module S2 and upper Inductive
module load
S1, and their mainAbout 7uH
difference is the current magnitude and
direction before turn on. The variation in the transient current is basically consistent after turning on.
direction before turn on. The variation in the transient current is basically consistent after turning
Choosing Δt = t1 − t2 = 100ns and recording the difference of the two moments, because the
5.3.
on. Indirect
Choosing ∆t = t1 − t2 = 100 ns and recording the difference of the two moments, because the
Measurement
oscilloscope has sampling error, tests are made many times for each testing method to compute the
oscilloscope
Due to has
the sampling
excellent error, tests
linearity of theare made many
transient currenttimes
of theforIGBT
eachmodule
testing at
method to compute
the turn-on the
arithmetic mean value to raise measurement accuracy. Table 5 shows the average valuestransient,
for fifty
arithmetic
the indirectmean value to raise
measurement takes measurement
the turn-on accuracy.asTable
transient 5 shows In
an example. theorder
average valuesthe
for fifty repeat
repeat measurements of the upper IGBT module and lower IGBT module atto verify
the turn-on consistency
transient.
measurements of the upper IGBT module and lower IGBT module at the turn-on transient.
of the transient current’s change rate under different testing conditions, a double pulse test is applied
on S1 and S2. At first, according to the phase leg testing circuit shown in Figure 5, S1 is tested with a

(a) (b)
Figure 17.
Figure 17. The
The current
current waveform
waveform at
at the
the turn-on
turn-on transient.
transient. (a) Second turning-on
(a) Second turning-on of S2; (b)
of S2; (b) Second
Second
turning-on of
turning-on of S1.
S1.

Table 5. Measured
Table 5. Measured current
current values.
values.

Δi(A) Δi1 Δi2 Δi3 Δi4

∆i(A) ∆i1 ∆i2 ∆i ∆i
S1 104.57 76.71 66.363 108.46 4
S1 S2 104.57
106.25 76.71
75.21 66.36
70.35 108.46
110.36
S2 106.25 75.21 70.35 110.36

According to Equations (4), (6), and Table 1, the partial stray inductance can be calculated, as
shown in Table 6. The difference of the two measured methods is within 5 nH, mainly deriving from
the sampling error of the oscilloscope.

Table 6. Calculated inductance based on Tables 1 and 3.

Energies 2017, 10, 1519 14 of 16

According to Equations (4), (6), and Table 1, the partial stray inductance can be calculated, as
shown in Table 6. The difference of the two measured methods is within 5 nH, mainly deriving from
the sampling error of the oscilloscope.

Table 6. Calculated inductance based on Tables 1 and 3.

nH Lσ 1 Lσ 2 Lσ 3 Lσ 4
Turn on transient of S1 31.45 47 40.68 28.11
Turn on transient of S2 30.69 47.9 39.8 24
Difference 0.76 0.9 0.88 4.11

5.4. Direct Measurement

The direct measurement is based on the inductance voltage integral method regardless of the
current linearity. Therefore, it can be realized at the turn-on transient or the turn-off transient.
The voltage sampling error is very small, because the voltage is based on digital integration function.
However, the current is sampled at t1 and t2 moments, so the current sampling error is relatively large.
In order to raise measurement precision, tests are made many times at both the turn-on transient and
the turn-off transient.
Due to the limited channel amounts of the oscilloscope, Lσ1 and Lσ2 are extracted firstly, as shown
in Figure 18a,b, in which the i1 and i2 are the current waveforms at the switching transient in branches
Energies 2017, 10, 1519 15 of 17
Lσ1 and Lσ2 . Also, uL1 and uL2 are the voltage waveforms at the switching transients on branches
L
and Lσ2. LItσ2is. It
σ1 and is observed
observed thatthat
the the voltage
voltage variation
variation is basically
is basically consistent
consistent at the
at the switching
switching transient.
transient. f1
fand
1 and f represent
f2 represent
2 the time integral of u
the time integral of uL1 andand u
L1 uL2 and and the integral time is 100
L2 the integral time is 100 ns. ns.

(a) (b)
Figure
Figure 18.
18. The
The test
test waveform
waveform at
at the
the turn-off
turn-off transient.
transient. (a)
(a) Turn-on
Turn-on transient;
transient; (b)
(b) Turn-off transient.
Turn-off transient.

The partial stray inductance for Lσ1 and Lσ2, based on the turn-on transient and the turn-off
The partial stray inductance for Lσ1 and Lσ2 , based on the turn-on transient and the turn-off
transient, are listed in Table 7. According to the same testing conditions, Lσ3 and Lσ4 are measured, as
transient, are listed in Table 7. According to the same testing conditions, Lσ3 and Lσ4 are measured, as
shown in Table 7. It is observed that the measurement difference is within 3 nH at both the turn-on
shown in Table 7. It is observed that the measurement difference is within 3 nH at both the turn-on
transient and the turn-off transient, so the turn-on transient and the turn-off transient are valid for
transient and the turn-off transient, so the turn-on transient and the turn-off transient are valid for the
the extraction of partial stray inductance.
extraction of partial stray inductance.
Table 7. The measured values of Lσ1 and Lσ2.
Table 7. The measured values of Lσ1 and Lσ2 .
nH Lσ1 Lσ2 Lσ3 Lσ4
nH Lσ 1
Turn-on transient 30.56 L σ 45.6
2 Lσ 3
39.56 26.6 Lσ 4
Turn-onTurn-off
transienttransient
30.56 31.37 45.6
42.77 41.37
39.56 27.57 26.6
Turn-off transient
Difference 31.37 0.81 42.77 2.83 41.37
1.81 0.97 27.57
Difference 0.81 2.83 1.81 0.97

The stray inductance mean values obtained from indirect measurement and direct measurement
are shown in Table 8. The measurement difference is within 5 nH so the consistency is very high.

Table 8. Comparison of stray inductance based on indirect and direct measurements.

nH Lσ1 Lσ2 Lσ3 Lσ4

Energies 2017, 10, 1519 15 of 16

The stray inductance mean values obtained from indirect measurement and direct measurement
are shown in Table 8. The measurement difference is within 5 nH so the consistency is very high.

Table 8. Comparison of stray inductance based on indirect and direct measurements.

nH Lσ1 Lσ2 Lσ3 Lσ4

Indirect measurement 31.56 47.4 40.24 26.01
Direct measurement 30.9 44.19 40.46 27.1
Difference 0.66 3.21 0.22 1.01

LσPc and LσNe are measured based on the proposed direct measuring method. As shown in
Table 9, the measurement values are almost the same as the simulation values and the calculated
values. So, simulation, calculation, and measurement can validate each other, which can further prove
the accuracy of the modeling and measuring method.

Table 9. Comparison of stray inductance based on simulation, calculation, and measurement.

Method LσPc LσNe

simulation 16.37 15.02
calculation 17.6 15.6
measurement 17.2 16.1

6. Conclusions
In this paper, the laminated bus-bar of a three-phase inverter in an electric vehicle with asymmetric
branching is investigated. For the first time, the imbalanced transient current on the bus-bar is deeply
analyzed from the view of stray inductance energy storage and release. The partial self-inductance
and mutual-inductance model of the asymmetrical parallel branches on the bus-bar is built and
the technique to reduce the equivalent inductance circuit models is developed based on matrix
calculation. Furthermore, a high-precision partial inductance measuring method considering spatial
electromagnetic transient field is proposed. At last, both the simulation values based on Q3D and
the calculation values based on matrix calculation are verified using the proposed direct and indirect
measurement methods.
For future work, frequency-domain simulations will be conducted to investigate the effects of
the asymmetrical stray inductance on the inverter’s EMI performance. This will provide insight for
effective component layout inside the enclosure of EV inverters.

Acknowledgments: The project is supported by the National Key Research and Development Program of China
(2016YFC0600906).
Author Contributions: Fengyou He and Chengfei Geng conceived and designed the experiments. Chengfei Geng
designed and performed the simulations. Jingwei Zhang performed the experiments. Chengfei Geng and
Hongsheng Hu wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Un-Noor, F.; Padmanaban, S.; Mihet-Popa, L.; Mollah, M.N.; Hossain, E. A Comprehensive Study of Key
Electric Vehicle (EV) Components, Technologies, Challenges, Impacts, and Future Direction of Development.
Energies 2017, 10, 1217. [CrossRef]
2. Bucher, A.; Schmidt, R.; Werner, R.; Leipenat, M.; Hasenohr, C.; Werner, T.; Schmitz, S.; Heitmann, A.
Design of a Full SiC Voltage Source Inverter for Electric Vehicle Applications. In Proceedings of the 18th
European Conference on Power Electronics and Applications (EPE’16 ECCE Europe), Karlsruhe, Germany,
5–9 September 2016; pp. 1–10.
Energies 2017, 10, 1519 16 of 16

3. Liu, J.; Sun, W.; Su, W.; Tai, X.; Wen, X. Inverter Output Power Density Increasing Technologies for EV/HEV
Applications. In Proceedings of the 19th International Conference on Electrical Machines and Systems
(ICEMS), Chiba, Japan, 13–16 November 2016; pp. 1–4.
4. Mühlfeld, O.; Paulsen, L.; Olesen, K.; Ströbel-Maier, H. Automotive Traction Inverter for Highest Power
Density. In Proceedings of the International Exhibition and Conference for Power Electronics, Intelligent
Motion, Renewable Energy and Energy Management, Nuremberg, Germany, 19–20 May 2015; pp. 871–878.
5. Caponet, C.; Profumo, D.; Doncker, W. Low stray inductance bus bar design and construction for good EMC
performance in power electronic circuit. IEEE Trans. Power Electron. 2002, 17, 225–231. [CrossRef]
6. Chen, C.; Pei, X.; Chen, Y.; Kang, Y. Investigation, Evaluation, and Optimization of Stray Inductance in
Laminated Busbar. IEEE Trans. Power Electron. 2014, 29, 3679–3693. [CrossRef]
7. Wang, J.; Yang, B.-J.; Xu, Z.-X.; Deng, Y.; Zhao, R.-X.; He, X.-N. Configuration of Low Inductive Laminated
Bus Bar in 750 kVA NPC Three-level Universal Converter Module of High Power Density. Proc. CSEE 2010,
30, 47–54.
8. Callegaro, A.D.; Guo, J.; Eull, M.; Gibson, J.; Preindl, M.; Bilgin, B. Bus Bar Design for High-Power Inverters.
IEEE Trans. Power Electron. 2017. [CrossRef]
9. Xuesong, W.; Zhengming, Z.; Liqiang, Y. Conducted EMI Reduction in IGBT Based Converters.
In Proceedings of the Asia-Pacific International Conference on Electromagnetic Compatibility (APEC),
Beijing, China, 12–16 April 2010.
10. Khan, M.; Magne, P.; Bilgin, B.; Wirasingha, S.; Emadi, A. Laminated Busbar Design Criteria in Power
Converters for Electrified Powertrain Applications. In Proceedings of the 2014 IEEE Transportation
Electrification Conference and Expo (ITEC), Dearborn, MI, USA, 15–18 June 2014; pp. 1–6.
11. Dong, Y.F.; Luo, H.Z.; Yang, H.Y.; He, Y.M.; Li, W.H.; He, X.G. Engineering Design for Structure and Bus Bar
of 1.2MV·A Hybrid Clamped Five-Level Converter Module. Trans. China Electrotech. Soc. 2016, 31, 11–18.
12. Schanen, J.L.; Clavel, E.; Roudet, J. Modeling of Low Inductive Bus bar Connection. IEEE Ind. Appl. Mag.
1996, 2, 39–43. [CrossRef]
13. Ruehli, A.E. Inductance Calculations in a Complex Integrated Circuit Environment. IBM J. Res. Dev. 1972,
16, 470–481. [CrossRef]
14. Paul, C.R. Inductance, Loop and Partial; Willey & Sons: New York, NY, USA, 2010; pp. 246–254.
15. Bryant, A.T.; Vadlapati, K.K.; Starkey, J.P.; Goldney, A.P.; Kandi-lidis, S.Y.; Hinchley, D.A. Current distribution
in high power laminated busbars. In Proceedings of the 14th European Conference on Power Electronics
and Application, Birmingham, UK, 30 August–1 September 2011; pp. 1–10.
16. Pasterczyk, R.J.; Martin, C.; Guichon, J.-M.; Schanen, J.-L. Planar Busbar Optimization Regarding Current
Sharing and Stray Inductance Minimization. In Proceedings of the European Conference on Power electronics
and applications, Dresden, Germany, 11–14 September 2005; pp. 1–9.
17. Zhang, N.; Wang, S.; Zhao, H. Develop Parasitic Inductance Model for the Planar Busbar of an IGBT H
Bridge in a Power Inverter. IEEE Trans. Power Electron. 2015, 30, 6924–6933. [CrossRef]
18. Eupec marketing department; European Power Semiconductor; Electronics Company. Measurement of the
Circuit Stray Inductance Lσ; Application Note; European Power Semiconductor and Electronics Company:
Warstein, Germany, 1999.
19. Zhu, H.; Hefner, R.; Allen, R.; Lai, J.-S. Characterization of power electronics system interconnect parasitics
using time domain reflectometry. IEEE Trans. Power Electron. 1999, 14, 622–628.
20. Feng, G.; Yuan, L.; Zhao, Z.; Zhao, J.; Lu, T. A Novel Stray Inductance Extraction Method for Bus Bars Based
on Turn-on/off Transient Process. Proc. CSEE 2014, 34, 6442–6449.
21. Wen, H.; Liu, J.; Zhang, X.; Wen, X. Electric Vehicle Drive Inverters Simulation Considering Parasitic
Parameters. In Proceedings of the 13th International Power Electronics and Motion Control Conference,
Poznan, Poland, 1–3 September 2008; pp. 417–421.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).