Download Designing a Three Phase Inverter for a Permanent Magnet Synchronous Motor

Lars Dautermann
Designing a Three Phase Inverter for a
Permanent Magnet Synchronous Motor
Helsinki Metropolia University of Applied Sciences
Bachelor of Science
Electronics
Thesis
June 19, 2013
Abstract
Author
Title
Lars Dautermann
Designing a Three Phase Inverter
for a Permanent Magnet Synchronous Motor
Number of Pages
Date
51 pages + 7 appendices
June 19, 2013
Degree
Bachelor of Science
Degree Programme
European Electrical Engineering
Specialisation option
Power electronics
Instructor
Arja Ristola, Technology Manager
The motivation for this thesis is to provide a three phase inverter for an electric car
in the Formula Student. Designing a inverter is a challenge that requires combined
knowledge in microcontroller programming, pcb layout, electric analysis and the
principals of how to control a motor.
The theory of electrical motors and especially permanent magnet synchronous
motors is shortly introduced. The physical and electrical characteristics were
studied to understand how the motor can be controlled with a three phase inverter.
The schematics, the design of drivers and safety features for the switching elements
in the inverter are shown.
The software to control the inverter and to communicate with other devices in a car
to obtain the information about the desired torque and power is explained.
A model of a space vector control in Matlab is simulated.
The advantages and disadvantages of different control methods like direct torque
control and field oriented control are discussed.
During the study of the thesis two inverter designs were designed. A low voltage, low
power version to study the concept and a high voltage, high power version for the
practical usage in the car. Due to the limitation of time only the low voltage inverter
is built and demonstrated.
Keywords
inverter, permanent magnet synchronous motor, mosfet, igbt
Contents
Nomenclature
1
1 Introduction
1.1 Formula Student . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Metropolia Motorsport . . . . . . . . . . . . . . . . . . . . . . . . .
3
3
3
2 Car System Overview
4
3 Theory
3.1 History of Synchronous Motors . . . . . . . . . . . . . . . . . . . .
3.2 Permanent Magnet Motor . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Physical Types of Permant Magnet Motors . . . . . . . . . .
3.2.2 Parameters of PMSM . . . . . . . . . . . . . . . . . . . . .
3.2.3 Dynamic Model of IPMSM - Flux Linkage . . . . . . . . . .
3.2.4 Dynamic Model of IPMSM - Stationary and Synchronous
Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.5 Torque Control . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1 Square Wave Conversion . . . . . . . . . . . . . . . . . . .
3.3.2 Modified Sine Wave Conversion . . . . . . . . . . . . . . .
3.3.3 Pure Sine Wave Conversion . . . . . . . . . . . . . . . . . .
3.3.4 Space Vector PWM . . . . . . . . . . . . . . . . . . . . . .
3.4 Control loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.1 Direct Torque Control . . . . . . . . . . . . . . . . . . . . .
3.4.2 Field Orientated Control . . . . . . . . . . . . . . . . . . . .
3.4.3 Clark-Park Transformation . . . . . . . . . . . . . . . . . . .
3.5 Switching Devices . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.1 MOSFET . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.2 IGBT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5
5
6
7
10
11
12
14
15
15
16
17
18
20
20
20
23
24
24
24
4 Matlab Simulation
27
5 Schematic
5.1 Power Supply Module . . . . . . . .
5.2 Microcontroller Modules . . . . . . .
5.3 Optocoupler Module . . . . . . . . .
5.4 CAN Bus Module . . . . . . . . . . .
5.5 Inverter Module . . . . . . . . . . . .
5.6 Resolver to Digital Converter Module
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
28
28
29
30
30
31
31
6 Inverter Parts
6.1 Switching Devices . . . . . . . . . . . . . . .
6.1.1 MOSFET Driver . . . . . . . . . . . .
6.1.1.1 Bootstrap Driver Calculation
6.1.2 IGBT Driver . . . . . . . . . . . . . . .
6.2 Protection . . . . . . . . . . . . . . . . . . . .
6.2.1 Short Circuit Protection . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
33
33
34
34
36
38
38
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6.2.2 RCD Snubber . . . . .
6.3 Microcontroller . . . . . . . .
6.4 Sensors . . . . . . . . . . . .
6.4.1 Temperature Sensors
6.4.2 Current Sensors . . .
6.4.3 Voltage Sensor . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
39
40
40
40
40
41
7 Measurements
7.1 Voltage Divider Resistance . . . . . . . . . . . . . . . . . . . . . .
7.2 Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
42
44
8 Program
46
9 Summary and Conclusion
47
List of Figures
48
List of Tables
50
Bibliography
51
A Matlab Code
A.1 Matlab code for square wave generation . . . .
A.2 Matlab code for modified sine wave generation
A.3 Matlab code for pure sine wave generation . .
A.4 Matlab code for space vector PWM . . . . . . .
52
52
53
54
56
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
Nomenclature
λa
Flux linkage phase a
λb
Flux linkage phase b
λc
Flux linkage phase c
λsd , λα
Flux linkage in stationary d-axis
λsq , λβ
Flux linkage in stationary q-axis
µ0
Permeability of the vacuum
ωr , ωe =
P
ω
2 r
Magnetic flux of the permanent magnets
ψm
θr , θ = θe =
Rotor speed
P
θ
2 r
Rotor flux angle
ea
Back EMF phase a
eb
Back EMF phase b
ec
Back EMF phase c
ia
Stationary current phase a
ib
Stationary current phase b
ic
Stationary current phase c
ied , id
D-axis synchronous current
isd , iα
D-axis stationary current
ieq , iq
Q-axis synchronous current
isq , iβ
Q-axis stationary current
kphase
Back EMF constant
Ld
D-axis synchronous inductance
Lq
Q-axis synchronous inductance
Lδ
Reluctance inductance component
Labcs
Inductance corresponding to the uniform air gap
Lrlc
Reluctance matrix
m
Space vector sector number
N
Number of turns of the d-axis winding
Pe
Electric power
2
Te
Torque
Ts
Switching period
vde , vd
D-axis synchronous voltage
vds , vα
D-axis stationary voltage
vqe , vq
Q-axis synchronous voltage
vqs , vβ
Q-axis stationary voltage
Vdc
Battery voltage
BLDC
Brushless direct current motor
CAN
Controller area network
CEMF
Counter electromotive eorce
f
Frequency
FOC
Field oriented control
I2C
Inter-integrated circuit
MOSFET
Metal oxide semiconductor field effect transistor
P
Number of motor poles
PM
Permanent magnet
PMSM
Permanent magnet synchronous motor
PWM
Pulse width modulation
SPI
Serial peripheral interface
SVPWM
Space vector pulse width modulation
3
1
Introduction
The motivation for this thesis was to provide a three phase inverter for an electric
car in the Formula Student. The Formula Student is an engineering competition
where students design and build a car under the rules of the Formula Student.
Points can be scored in different engineering and business events throughout
the competition. Writing this thesis and creating a three phase inverter helps to
understand the electric car in more detail and gain points in the design event. A
self built and self designed inverter in the car can be better tailored for the desired
purposes and also helps to assess electromagnetic interference problems better
that might occur in the car’s communication network.
This study is carried out for the Metropolia Motorsport team.
1.1
Formula Student
The Formula Student is an engineering competition that took place the first time in the USA
in 1981. The American competition is the Formula SAE (Society of Automotive Engineers ).
In 2013 there are 11 official competitions internationally: three in the USA, one in Australia,
Figure 1: Formula Student logos
one Brazil, one in Italy, one in the United Kingdom, one in Germany, one in Austria, one in Hungary and one in Japan.[6, p. 6]
[2] The Formula Student in these countries use the same basic set of rules for
its competitions which are slightly changed depending on the event. Teams are
created by students to represent their universities. To the right the logo of the
overall formula student and the logo of the German competition can be seen.
1.2
Metropolia Motorsport
The Metropolia Motosport team was formerly
known as Helsinki Polytechnic Formula Engineering Team, which was founded in 2000.
The first competition participation was in 2002
in Birmingham UK with the car HPF002. In Figure 2: Metropolia Motorsport
2008 the team changed its name to Metropo- logo
lia Motorsport due to the name change of the
university to Helsinki Metropolia University of Applied Sciences. Since then the
team has constructed a car each year and participated in several competitions.
4
In 2013 the team constructed its first fully electric car. In cooperation with ABB
they built their own permanent magnet synchronous motor, which is controlled by
a inverter from ABB.[1]
In figure 2 the logo of the Metropolia Motorsport team can be seen.
2
Car System Overview
Controller
Inverter 1
Motor 1
Inverter 2
Motor 2
DC Bank
Dashboard
Driver Information
Pedals
Torque Control
Input
Drive Train
Figure 3: System Overview
In figure 3 the top view of the electric car can be seen. The drive train is located in the rear of the car behind the driver’s seat. At the front of the car the
pedals for acceleration and braking can be found.
The car has a lot of electrical devices and sensors on board. To control all those
devices a central control unit is used which runs a self developed software. This
controller is also called main control unit (MCU). The separate electrical systems
in the car communicate by using a CAN bus. The CAN bus was specifically created to guarantee stable communication in automotive environments. It uses a
differential signal. This helps to prevent coupling of electromagnetic interferences.
For the driver to be able to control the car, the input devices have to be transformed into electrical signals. The two front pedals positions are sensed using a
high precision linear resistor. The steering wheel has an encoder implemented
which reveals the momentary angle. These signals are evaluated by the main
5
controller. According to the software the acceleration and steering angle information is calculated into torque and speed demands for the rear wheels. Via the
CAN bus this information is send to the inverters in the drive train. The inverters
use the received information to each separately control one motor.
Before the car can be set into ready to drive mode a series of safety procedures
have to be executed by the driver.
All systems are constantly monitored for failures by software implemented routines as well as hardware implemented safety features.
3
3.1
Theory
History of Synchronous Motors
Synchronous machines were first development during the middle of the 19th century as single phase generators to supply lighting systems. In the early development the salient pole rotor motor and the turbo generator formed as the typical
design type. Being closely connected with the development of power plants the
usage and size of synchronous motors grew with the size of the power plants. In
an industrial manufacturing environment the synchronous motor was used where
a constant speed or usage as a phase shifter was needed.
With the development of frequency inverters the synchronous motor can now
be used as a speed controlled motor and only recently found its way to new applications ranging from smallest sizes, like watches, phonographs and precision
mechanics up to the biggest sizes and powers used in cement mills, conveyors
and steel mills. [3, p. 289]
The trend is to use high efficiency motors like Permanent Magnet Synchronous
Motors (PMSM) in home applications such as refrigerators, air conditioners and
washing machines.
A nice side effect is that the efficiency of permanent electric motors can play
an important role to reduce greenhouse gases and help reduce the air pollution
caused by exhausts. The requirement of a clean environment and the reduction
of a petroleum dependency has lead to a a paradigm shift. Slowly gasoline engines are replaced or at least assisted by electrical machines. In vehicle power
trains more and more vehicles are using hybrid or fully electrical systems. Ongoing progress in CPU and power semi conductor performance has made it possible
to easily implement complicated control techniques and algorithms using cost effective components. [4, p. xi]
6
3.2
Permanent Magnet Motor
As seen in Figure 4 AC motors can be divided into asynchronous and synchronous
motors. Like any AC motor asynchronous motors work on the principle of creating a rotating field. [3, p. 170]. Technically asynchronous motors resemble transformers. The stator being the primary and the rotor being the secondary side.
Asynchronous motors are called asynchronous because the speed of the rotor is
less than the speed of the rotating field. This behaviour is called slip.
Synchronous motors use the stator’s and rotor’s magnetic field to lock with
each other in respect to their position leading to a synchronous rotation of the
rotor with the stator’s magnetic field. The speed of a synchronous motor is n =
120∗f
. If a motor had 4 poles and was supplied with a 50Hz frequency the speed
P
would be 1500rpm.
Reluctance motors use the effect of attraction experienced when holding a
magnet to a magnetisable material like an iron nail. The rotor consisting of a
magnetisable material is attracted to the magnetic field created by the stator’s
coils.
Permanent magnet motors use the effect of repulsion of equally magnetic polarized fields experienced when bringing two magnets together with the same
pole. The rotor’s magnets are being pushed away by the magnetic field created
in the stator coils.
AC-Motors
Asynchronous
Squirrel Cage
Induction Motor
Synchronous
Doubly Fed
Induction Motor
BLDC Motor
Permanent
Magnet
Motor
Reluctance
Motor
PMSM
Surface
PM
Motor
Switched
Reluctance
Synchronous
Reluctance
Interior
PM
Motor
Figure 4: Taxonomy of AC motors
Permanent magnet motors are classified into two categories based on the
shape of their counter clectromotive force (CEMF). According to Faraday’s law
RR
ΦB =P B(r, t)dA, CEMF is created when a coil rotates in a magnetic field or the
(t)
7
other way round a static coil experiences a rotating magnetic field. The change
in magnetic flux then induces a voltage according to eb = Kb ωr , where Kb is the
back EMF constant, and ωr is the rotor angular speed. [4, p. 2]
Counter EMF PMSM
e
e
e
c
e
a
e
b
e
c
Voltage[V]
b
Voltage[V]
a
Counter EMF BLDC
Time [s]
Time [s]
(a)
(b)
Figure 5: (a) Continuous Counter EMF of a PMSM, (b) Trapezoidal Counter EMF
of a BLDC
Because the thesis concentrates on a inverter used for a PMSM this type of
permanent magnet motor will be focused on.
3.2.1
Physical Types of Permant Magnet Motors
In Figure 6 typical PMSM structures are shown. Motors where the PMs are attached to the surface of the rotor are called surface PMSM (SPMSM). Motors
where the PMs are inserted into the rotor are called interior PMSM (IPMSM). The
different arrangement of the permanent magnets leads to different reluctances
and inductances. Because modern PMs have a permeability close to 1 they can
be treated like air regarding the magnetic reluctance. PMs embedded in the rotor
almost always lead to the q-axis synchronous inductance being larger than the
d-axis synchronous inductance,
Lq > Ld ,
with
Ld =
µ0 N 2 A
2(g + hm )
and
Lq =
µ0 N 2 A
.
2g
8
Figure 6: Typical PMSM structures: (a) surface magnet, (b) inset magnet, (c) pole
shoe rotor, (d) tangentially embedded magnets, (e) radially embedded magnets
(flux-concentration), (f) two magnets per pole in the V position, (g) a synchronous
reluctance motor equipped with permanent magnets.[8, p. 397]
while PMs on the rotor surface lead to the q-axis synchronous inductance
being equal to the d-axis synchronous inductance.
Lq = Ld ,
with
Ld = Lq =
µ0 N 2 A
.
2(g + hm )
With the arrangement in Figure 6(e) a flux concentration can be achieved, for
example an effective flux density of 0,8T using magnets with 0,4T flux density.
Though the use of PM mounted on the surface makes better use of the PMs flux,
because the magnets in SPMSMs are mounted on the surface the speed range
and power density is smaller compared to IPMSMs due to the centrifugal force
that acts upon them and magnet flux leakage, which can even lead to demagnatisation of surface mounted PMs due to the heat created by the losses. [4, p.
140],[8]
Figure 7 shows the physical structure of the stator and rotor of the motor from
the Metropolia Motorsport team. It has 24 slots in the stator and 4 magnetic poles
on the rotor. The magnets are tangentially embedded. This design was chosen
to allow speed up to 6000 rpm and and efficient use of power of 8kW-22kW.
9
S lo t
Stator
North
Rotor
South
South
Slot
Slot
Rotor
Slot
North
Figure 7: Structure of the Metropolia Motorsport’s motor: interior magnet
10
3.2.2
Parameters of PMSM
Given the physical description of a PMSM the most important parameters to describe the motor mathematically are shown in table 1.
Physical
tion
representa-
Parameter representation
Dimension
Resistance of the motor’s stator winding
RS
Ω
D-axis inductance of one
motor phase. This is
a transformed value derived by the Park transformation.
Ld
H
Q-axis inductance of
one motor phase. This
is a transformed value
derived by the Park
transformation.
Lq
H
Back EMF constant.
Kp
Vs
rad
Rotor Pole Pair.
P
1
Table 1: PMSM parameter overview
11
3.2.3
Dynamic Model of IPMSM - Flux Linkage
A simplified equivalent circuit of any AC machine can be seen in Figure 8.
The inductance can be divided into stator and rotor inductance. The inductance can be described as [4, p. 149]
Labcs = Labcs − Lrlc (θ),
where


cos 2θ
cos(2θ − 2π/3) cos(2θ + 2π/3)


Lrlc (θ) = Lδ cos(2θ − 2π/3) cos(2θ + 2π/3)
cos 2θ
,
cos(2θ + 2π/3)
cos 2θ
cos(2θ − 2π/3)
with
N 2A
γ2 .
2
Lδ describes the reluctance component.
θ is the rotor angle.
µ0 is the permeability of the vacuum.
A is the air gap area through which the flux crosses.
N is the number of turns of the d-axis winding.
γ2 is a positive constant describing the air gap.
Lrlc is called reluctance matrix and varies along with the rotor angle.
Lδ ≡ µ0
L
ia
va
ea
vc
ec
ib
vb
L
eb
ib
L
Figure 8: Simplified equivalent circuit of a 3 Phase AC motor.
12
Labcs is the inductance corresponding to the uniform air gap described as

Labcs

Lms + Lls −1/2Lms −1/2Lms


=  −1/2Lms Lms + Lls −1/2Lms  ,
−1/2Lms −1/2Lms Lms + Lls
with
Lms ≡ µ0
N 2A
γ0 .
2
γ0 and γ2 are positive constants.
The total flux linkage is
 
 


λa
ia
cos θ
 
 


 λb  = [Labcs − Lrlc (θ)]  ib  + ψm cos(θ − 2π/3 .
λc
ic
cos(θ + 2π/3
Figure 9: Stationary equivalent circuit of a PMSM (one phase)
3.2.4
Dynamic Model of IPMSM - Stationary and Synchronous Frame
The flux linkage in (3.2.3) is transformed into the stationary d,q frame as
" # "
#" #
"
#
λsd
Ls − 3/2Lδ cos 2θ
−3/2Lδ sin 2θ
isd
cos θ
=
+ ψm
.
λsq
−3/2Lδ sin 2θ
Ls + 3/2Lδ cos 2θ isq
sin θ
From this the stationary IPMSM dynamic model is obtained through
d s
λ
dt dq
An overview of the dynamic models of SPMSM and IPMSM can be seen in
Table 2.
s
vdq
= rs isdq +
13
SPMSM in the stationary frame
d
dt
s
id
=
isq
rs
Ls
s
id
−
isq
ψm ωe
Ls
− sin θe
+
cos θe
1
Ls
s
vd
vqs
IPMSM in the stationary frame
s
s s
vd
id
Ls − 32 Lδ cos 2θe
− 32 Lδ sin 2θe
d id
= rs s +
dt is
vqs
iq − 32 Lδ sin 2θe Ls + 32 Lδ cos
2θ
e q
− sin 2θe cos 2θe isd
− sin θe
−3ωe Lδ
+ ωe ψm
cos 2θe sin 2θe isq
cos θe
SPMSM in the synchronous frame
d
dt
e −rs
e
ω
id
id
e
L
s
−
rs
e =
iq
−ωe − Ls ieq
ψωe
Ls
0
+
1
1
Ls
e
vd
vqe
IPMSM in the synchronous frame
d
dt
#
e " −rs
Lq e ω
id
id
e
Ld
Ld
−
=
Ld
rs
ieq
ieq
−ωe Lq − Ls
ψωe
Lq
" 1 e#
v
0
+ L1d de
v
1
Lq q
Table 2: PMSM dynamic equations[4, p. 154]
3
Ld = Ls − Lδ
2
3
Lq = Ls + Lδ
2
3
Ls = Lms + Lls
2
14
3.2.5
Torque Control
Torque in the rotating frame (synchronous) is calculated as:
Te =
3P ψm ieq − (Lq − Ld )ied ieq .
4
ψm ieq is the electro-magnetic torque based on the Lorentz force, which is build
by the quadrature current.
−(Lq − Ld )ied ieq is the reluctance torque caused by the Ld − Lq asymmetry.
Using a lossless model where vde = −ωe Lq ieq and vqe = ωe Ld ied + ωe ψm the total
electric power is
3P
ωr ψm ieq + (Ld − Lq )ied ieq .
4
From this the torque equation in the stationary frame can be derived as
Pe =
Te =
3P s s
(λ i − λsqs isd ).
4 ds q
where
3
λsd = Ls isd − Lδ (−isd cos 2θe + isq sin 2θe ) + ψm cos θe ,
2
3
λsq = Ls isq − Lδ (isd sin 2θe − isq cos 2θe ) + ψm sin θe .
2
15
3.3
Inverter
There are several methods to convert DC to AC. They differ mainly in their approximation to a perfect sinusoidal signal. As one would expect the best approximation
yields the best transfer of power to a device that expects a sinusoidal signal. A
better approximation comes with more complexity and cost. A compromise between functionality, cost and complexity has to be made.
In this part the advantages and disadvantages of different conversion methods
will be discussed and evaluated.
3.3.1
Square Wave Conversion
The simplest form of DC to AC conversion is a square wave. As shown in figure 10
it uses the full DC voltage and switches between this maximum level at 50% duty
cycle. This hard switching could be used for a BLDC as this type of motor expects
a DC signal. For a PMSM it is not desirable because it expects a sinusoidal
signal. The advantages of square wave conversion are its simplicity and low cost
of components. The disadvantages are high amplitude harmonics (Figure 11)
and unsmooth behaviour of devices that expect a sinusoidal signal, which leads
to physical stress.
SquareWave
1.5
Voltage[V]
1
0.5
0
−0.5
−1
0
0.2
0.4
0.6
0.8
1
Time [s]
1.2
1.4
1.6
1.8
2
Figure 10: Square wave conversion
The Matlab code to create figure 10 and 11 can be seen in appendix A.1.
16
Amplitude Spectrum of SquareWave
1
|Voltage[V]|
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
Frequency [Hz]
7
8
9
10
Figure 11: Square wave amplitude spectrum
3.3.2
Modified Sine Wave Conversion
In figure 12 a modified sine wave conversion can be seen. It is an improvement
in comparison to the square wave conversion. Instead of just two levels and a 50
% duty cycle it uses three levels. The
Modified Sinus Wave
1
Voltage[V]
0.5
0
−0.5
−1
0
0.2
0.4
0.6
0.8
1
Time [s]
1.2
1.4
1.6
1.8
2
Figure 12: Modified sine wave conversion
The Matlab code to create figure 12 and 13 can be seen in appendix A.2.
17
Amplitude Spectrum of Modified Sine Wave
1
|Voltage[V]|
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
Frequency [Hz]
7
8
9
10
Figure 13: Modified sine wave amplitude spectrum
3.3.3
Pure Sine Wave Conversion
In a pure sine wave conversion, the desired sinusoidal signal is compared against
a sawtooth wave. Whenever the desired sinusoidal signal and the desired value
intersects with the sawtooth sampling frequency a new pulse is started until the
sawtooth signal reaches the maximum amplitude. A Matlab generated example
can be seen in figure 14.
Pure Sine Wave PWM switching
1
Vdc
Vdc/2
|Voltage[V]|
0.5
0
−0.5
−1
0
0.1
0.2
0.3
0.4
0.5
Time [s]
0.6
0.7
0.8
0.9
1
Figure 14: Simulated PWM signal for a pure sine wave
In figure 15 the amplitude spectrum of the pure sine wave PWM signal can be
seen. It has components spread across a wide range of frequencies because of
the changing duty cycles during the modulation.
The Matlab code to create figure 14 and 15 can be seen in appendix A.3.
18
Amplitude Spectrum of sine wave
1
|Voltage[V]|
0.8
0.6
0.4
0.2
0
0
10
20
30
40
50
60
Frequency [Hz]
70
80
90
100
Figure 15: Simulated PWM signal for a pure sine wave amplitude spectrum
3.3.4
Space Vector PWM
In space vector PWM the different switching states of the inverter are mapped into
a hexagon. The state vectors are placed analog to the abc frame. By doing so the
synchronous voltage vectors can be obtained by their amplitude and angle. When
they fall into the designated sector a sequence of vector states is run to synthesize
the desired output voltage. To reduce the computing that has to be done the
values from the stationary abc frame are transformed into the dq synchronous
frame. The voltages vde and vqe are calculated in the rotating frame as:
e
e
vde = (P I)(ie∗
d − id ) − ωLq iq ,
(3.1)
e
e
vqe = (P I)(ie∗
q − iq ) − ωLd id + ωψm .
(3.2)
These values are then transformed back into the stationary α,β frame as seen
in figure 16. The maximum rms voltages in space vector PWM that can be
achieved are higher than with sinusoidal PWM. This is because of the utilisation of the 3rd harmonic.
√
The maximum rms voltage for sinusoidal PWM is 2·√32 · Vdc ≈ 0.612 · Vdc .
The maximum rms voltage for space vector PWM is √12 · Vdc ≈ 0.707 · Vdc
Because space vector PWM utilizes a higher voltage, more or equal power can
be transferred using less current.
In figure 17 it can be seen how the voltage vector is synthesized. The desired
voltage vector is Va . The vector V1 and V2 represent the configuration of the upper
switches of the inverter. To create vector Va the switch configuration V1 is switched
on for a proportional time T1 of Va1 . Then V2 is switched on for a proportional time
T2 of Va2 . The duration of each switch position is calculated from vα and vβ as: [4,
p. 278]
19
β
β
b
V3 (0, 1, 0)
V2 (1, 1, 0)
Sector 2
a
Sector 3
α
Sector 1
V0 (0, 0, 0)
V4 (0, 1, 1)
V7 (1, 1, 1)
Sector 4
Sector 6
α
V1 (1, 0, 0)
Sector 5
V5 (0, 0, 1)
c
(a)
V6 (1, 0, 1)
(b)
Figure 16: Stationary frame of the three phases abc with overlayed stationary α,β
frame (a), Space dector diagram (b)
√
3 · Ts
T1 =
V
√ dc
3 · Ts
T2 =
Vdc
h
i
πm
πm
· sin(
· vα − cos(
· vβ
3
3
π(m − 1)
π(m − 1)
· − sin(
· vα − cos(
· vβ .
3
3
T0 = Ts − (T1 + T2 );
Ts is the the switching period.
T0 is the zero vector switching time.
m is the number of the sector.
Vdc is the battery voltage.
β
V2 (1, 1, 0)
Va2
Va
Sector 1
Va1
α
V1 (1, 0, 0)
Figure 17: Space vector synthesizing
20
3.4
Control loop
To control the speed and torque of a PMSM the frequency and voltage have to
be controlled. Thus the devices to control such a motor are called variable frequency drive (VFD). Mainly two different control types are distinguished: scalar
control and vector control. Scalar control uses the effect of changing the operating parameters such as voltage and frequency to drive the motor at different sets
of characteristic curve families. It is called scalar control because it scales the
operating parameters. [9, p. 671]
Vector control dynamically controls the motor’s torque directly by observing the
motor’s current and keeping the magnitude of the instantaneous magnetizing current space vector constant so that the rotor flux linkage remains constant.[9, p.
645] It is called vector control because it analysis voltage, current and magnetic
flux based on their vector representation. Two types of vector control, direct
torque control (DTC) and field oriented control (FOC), are discussed in more
detail in the following section.
3.4.1
Direct Torque Control
Direct torque control uses the measurement of voltage and current to calculate
and estimate the magnetic flux and torque. It can be divided into direct self-control
(DSC) and space vector modulation (SVM).
3.4.2
Field Orientated Control
The overall control loop looks similar to figure 18. The torque is controlled by the
quadrature current (iq ) of the rotor. The current component id does not need to
be controlled. id is normally used to create the flux of the excitation circuit. In a
PMSM this flux is constantly present because of the magnets. id can be used to
induce field weakening.
As seen in figure 18 the three phase current are measured with an ADC (analog
to digital inverter) of a microcontroller. This information is then transformed from
the stationary three phase stator frame into the rotating rotor frame. Further this
information is subtracted from the desired current. A PI controller regulates how
much the current has to be increased or decreased. This is then calculated into
the corresponding voltage values. After an inverse park transformation from the
rotating dq frame into the stationary αβ frame the appropriate switching sequence
is executed.
21
Figure 18: Control loop overview
In a more compact way it can be said that field oriented control is the technique used to achieve the decoupled control of torque and flux by transforming
the stator current quantities (phase currents) from the stationary reference frame
to torque and flux producing currents components in the rotating (synchronous)
reference frame(Figure 19).[7]
A comparison overview between DTC and FOC can be seen in table 3 [10].
22
Comparison property
DTC
FOC
Dynamic response to
torque
Coordinates reference
frame
Low speed (¡ 5% of
nominal) behavior
Very fast
Fast
alpha, beta (stator)
d, q (rotor)
Requires speed sensor
for continuous braking
Controlled variables
torque & stator flux
Good with position or
speed sensor
rotor flux, torque current
iq & rotor flux current id
vector components
Steady-state torque/current/flux ripple & distortion
Parameter
sensitivity,
sensorless
Low (requires high quality current sensors)
Low
Stator resistance
Parameter
closed-loop
d, q inductances, flux
(near zero speed only)
d, q inductances, rotor
resistance
d, q inductances, rotor
resistance
Required (either sensor
or estimation)
sensitivity,
Rotor position measurement
Not required
Current control
Not required
Required
PWM modulator
Not required
Required
Coordinate transformations
Switching frequency
Not required
Required
Varies widely around average frequency
Lower (requires high
quality current sensors)
spread spectrum sizzling noise
Constant
Switching losses
Audible noise
Control tuning loops
speed (PID control)
Low
constant
frequency
whistling noise
speed (PID control), rotor flux control (PI), id
and iq current controls
(PI)
Complexity/processing
Lower
Higher
requirements
Typical control cycle 10-30 microseconds
100-500 microseconds
time
Table 3: Comparison between DTC and FOC
23
3.4.3
Clark-Park Transformation
Magnet Synchronous Motor
ThePermanent
Clark-Park
transformation combines two steps. The Clark transformation and
the Park transformation.
With the Clark transformation the stationary 3 phase
Transformations
Iq and Id are referenced
the rotating reference
frame. Hence ithe
measured
currents ia , iInb FOC,
andtheiccomponents
are transformed
into thetostationary
components
iβ . In
α and
stator currents have to be transformed from the three phase time variant stator reference frame to the
figure 19 thistwofirst
step dq
is rotor
visualized.
axis rotating
reference frame. This can be done in two steps as shown in Figure 1-5.
ia
iD
Z
Z
id
F
Z
F
q-axis
F
ib
ic
Clarke Transformation
iq
iE
d-axis
Park Transformation
Figure 1-5 • Forward Transformations
Figure 19: Clarke and Park transformation [7, p. 9]
The transformation from the 3-phase 120 degree reference frame to two axis orthogonal reference frame
is known as Clarke transform. Similarly the transformation from two axis orthogonal reference frame to
the two axis rotating reference frame is known as Park transform.
The Clark transformation is mathematically described as:
Clarke Transformation
 
" #
"
# ia
The measured motor currents are first translated
reference frame to the two axis
1 from the
iα The transform
− 12 by3-phase
2 1 −
 following

orthogonal reference frame.
is
the
equations.
2
√expressed
√
=
i


b
3 0 23 − 23
Iα = Ia
iβ
ic
Iβ = (Ia + 2Ib)/√3
Where Ia+Ib+Ic = 0
In the next step the park transformation is applied. Here the stationary comParkiβTransformation
ponents iα and
are transformed into the rotating (synchronous) components id
The two axis orthogonal stationary reference frame quantities are then transformed to rotating reference
and id . In figure
19 this second step is visualized.
frame quantities. The transform is expressed by the following equations:
Id = Iα cosθ + Iβ sinθis mathematically described as:
The Park transformation
Iq = Iβ cosθ – Iα sinθ
" # r "
#" #
i
cos
θ
sin
θ
iα
2
d
Inverse Park Transformation
=
3 provide
− sin
coscomponents
θ iβ in the rotating reference frame.
Now the outputs of theiqPI controllers
theθvoltage
These
Thus an inverse of the previous process has to be applied to get the reference voltage waveforms in
stationary reference frame. At first, the quantities in rotating reference frame are transformed to two axis
orthogonal
reference
frameinto
usingone
Inverse
Park transformation.
Park
two
stepsstationary
can be
formed
step
known as The
dq0Inverse
transformation:
transformation is expressed by the following equations:
Vα = Vd cosθ - Vq sinθ
 
# ia
cos θ − 2π
sin θ + 2π
 
3
3
 ib 
2π
2π
3 − sin θ − sin θ − 3
− sin θ + 3
ic
" =#
r+ Vd"sinθ
Vβ
Vq cosθ
id
2 cos θ
iq
=
To regain the stationary three phase components reverse transformations can
be applied. In FOC it is necessary to inverse transform the stationary voltage
components
vd and vq into the stationary
vα and vβ .
10
R ev is icomponents
on 0
" # "
#" #
vα
cos θ − sin θ vd
=
vβ
sinθ cosθ
vq
given in terms of the device data sheet parameters Ciss , Coss ,
and Crss as follows:
Cgd ¼ Crss
Cgs ¼ Ciss ÿ Crss
power supplies and brushless dc motor driv
large conduction losses, its power rating is
kilowatts. Because of its many advantages o
24have received hig
modern MOSFET devices
tance.
Cds ¼ Coss ÿ Crss
3.5
Switching where
Devices
Crss is the small-signal reverse transfer capacitance; Ciss
6.6 MOSFET Regions of Oper
is the small-signal input capacitance with the drain and source
terminals shorted; and Coss is the small-signal output capaci- Most MOSFET devices used in power electro
3.5.1 MOSFET tance
with the gate and source terminals shorted.
are of the n-channel, enhancement type, lik
The MOSFET capacitances Cgs , Cgd and Cds are nonlinear Fig. 6.6a. For the MOSFET to carry drain c
and a function of the dc bias voltage. The
variations in transistor.
Coss between the
Mosfet stands for metal-oxide-semiconductor
field-effect
The
drainmosfet
and the source must be cre
and Ciss are signi®cant as the drain-to-source voltage and the when the gate-to-source voltage exceeds the
nowadays has replaced
the bipolar
transistor
low
power
switching
gate-to-source
voltage each
cross zero.inThe
objective
of the
voltage Vapplications
Th . For vGS > VTh, the device can
drive circuit is to charge and discharge the gate-to-source and triode region, which is also called ‘‘constant r
almost completely. Being
a voltage
is off
more
and faster.
gate-to-drain
parasitic controlled
capacitances todevice
turn on it
and
the efficient
or in the saturation region, depending on the
device, respectively.
given v , with small v (v < v ÿ V )
GS
Cgd
Cds
Cgs
(a)
G
S
S
SiO2
n+
Cds
p
n+
Cgs
Cg
d
DS
DS
GS
Th
ates in the triode region (saturation region in
larger vDS ðvDS > vGS ÿ VTh ), the device ent
region (active region in the BJT). For vGS <
turns off, with drain current almost equal to
regions of operation, the gate current is alm
why the MOSFET is known as a voltage-d
therefore, requires simple gate control circui
The characteristic curves in Fig. 6.6b sho
three distinct regions of operation labeled
saturation region, and cut-off region. When
ing device, only triode and cut-off regions a
when it is used as an ampli®er, the MOSFET
the saturation region, which corresponds to
in the BJT.
The device operates in the cut-off region
vGS < vTh , resulting in no induced channel. In
the MOSFET in either the triode or satu
channel must ®rst be induced. This can be
applying gate-to-source voltage that exceeds
n-
vGS > VTh
n+
Once the channel is induced, the MOSFET
either the triode region (when the channel is
no pinch-off, resulting in drain current pro
D
channel resistance) or the saturation regi
(b)
pinches off, resulting in constant ID ). The g
FIGURE 6.9 (a) Equivalent MOSFET representation including junc- voltage (vGD ) determines whether the induc
tion capacitances;
and (b) representation
of this physical location.
or not.capaciThis is subject to the follo
Equivalent
MOSFET
representation
includingpinch-off
junction
Figure 20: (a)
tances; and (b) representation of this physical location.[9, p. 83]
In figure 20 you can see the physical structure of a mosfet with the location of
the capacitances (b) and an equivalent representation (a). The behaviour of the
capacitance charging and discharging is almost exactly the same as with an igbt.
In the following section this is explained in more detail. To understand the similar
behaviour in can be assumed that the drain of a mosfet is the collector of an igbt
and the source of a mosfet is the emitter of an igbt.
3.5.2
IGBT
Igbt stands for insulated-gate bipolar transistor. The igbt combines the advantages of a mosfet and a bipolar transistor. The gate of an igbt, like a mosfet, is
voltage driven and needs little to ideally zero gate current. Like a bipolar transistor
it is capable of conducting high currents. In figure 21 the physical structure, the
equivalent circuit and the capacitance model are shown.
driver
outputbetween
performance
forprevents
switching
mited
interaction
biased, which
any current ¯ow andrequirements
the device is in its for an IGBT driver circuit, the
ctronic circuit
designers.
rmation
given
in this
note
is the gate charge. This gate cha
reverseapplication
blocking state. If
the gate terminalparameter
is kept at ground
ce
reliability
are
observed
potential
but
a
positive
potential
is
applied
to
the
collector,
the
ly and does not constitute complete
characterised by the equivalent input 25
capacitance
ÿ
tual circuits. This signi®- pn-junction between the p-base and n -drift region is reverseepower
information
is not exhaustive. The
and CGEis. in its
electronic system biased. This prevents any current ¯ow and the device
proper
remains with the user.
ment
timedesign
can be signi®e performance and reliaIGBT Capacitances
the design stage. As high
n circuit applications, it is
ent to model the IGBT
s. However, development
nent of every power electronic system is
ower modules themselves – the IGBT
ms the vital interface between the power
e controller. For this reason, the choice
the calculation of the right driver output
y linked with the degree of reliability of a
n. Insufficient driver power or the wrong
(a)
r may result in module and
driver(a)
ation of MOSFET and BJT.
(b)
(b)
FIGURE 7.2 The IGBT (a) half-cell vertical cross section and (b)
equivalent 21:
circuit The
model.IGBT (a) half-cell vertical cross
Figure
section and equivalent circuit
model[9, p. 102], (b) IGBT capacitances
The following table explains the designation
capacitances. In IGBT data sheets these capaci
arecircuit
specified
voltage-dependent
low
Curve
Take a look at the igbt equivalent
in figure 21 as
(a). The
igbt works as if
capacitances
of
IGBTs
in
the
“off”
state.
a bipolar npn transistor’s gate was driven by a mosfet’s gate. This npn transistor
capacitances are independent of temperature
haviour (turn-on andbecomes
turn-off)conductive
of an IGBT
and drives the gate of the high power pnp transistor.
dependent on the collector-emitter voltage, as sh
ermined by its structural, internal
To switch an igbt efficiently the gate
needs
to be driven
into saturation is substa
the mosfet
following
curve.
This dependency
arges) and the internal and outer
as quickly as possible. Thereforehigher
the capacitances
seen
in figure 21 (b) voltage.
have
at a very low
collector-emitter
N
to be charged. In figure 22 (a) the sequence of charging the capacitances is
shown. At first the gate-emitter capacitor is charged. When the gate-emitter
voltage VGE(th) is reached, the igbt begins to conduct. When VGE(pl) is reached the
gate-collector capacitance is charged in parallel. After crossing the Miller plateau
the gate-emitter capacitance becomes fully charged an VGE(on) is reached. The
2007-10-31 – Rev00
gate charge of an igbt plays an important role in selecting and calculating a driver.
The information about the gate charge can be found in the datasheet of the igbt
manufacturer. In 22 (b) an example gate charge curve can be seen. Depending
on the voltage applied to the collector and emitter and the ambient temperature,
the capacitance varies.
capacitance,
IGBT
saturation.charging of the gate-collector capacitance
and charging of the gate-emitter capacitance until full
saturation.
To IGBT
calculate
the switching behaviour and the driver, the
input capacitances may only be applied to a certain
To calculate the switching behaviour and the driver, the
input capacitances may only be applied to a certain
The
charge
increases
in to
linea with
the extent.
current At
rating
the gate
DC-Link
voltage,
albeit
lesser
higher
of
IGBT modules.
The
charge
is also
dependent
operation
voltages
thegate
gate
charge
increases
due on
to the
the
DC-Link
voltage,ofalbeit
a lesser
extent. At higher
larger
influence
the toMiller
capacitance.
In most
26
operation
voltages
the gate
charge increases due to the
applications
this effect
is negligible.
larger influence of the Miller capacitance. In most
applications this effect is negligible.
Simplified Gate Charge Waveforms
Gate Charge Characteristic
Simplified Gate Charge Waveforms
Gate Charge Characteristic
VGE(on)
VGE(on)
GE(off)
VV
GE(off)
QGE
QGE
(a)
QGCQGC
QGEQGE
(b)
t0 switching
interval:The
Thegate
gatecurrent
currentIGIG charges
charges the
the input
input capacitance
voltage
VGEVrises
to to
t0 switching
interval:
capacitance CCGEGEand
andthe
thegate-emitter
gate-emitter
voltage
GE rises
. Depending
onthe
the gate
gate resistor,
resistor,
several
may
in in
this
state.
As As
VGEVis
still
VGE(th)
Figure 22: (a) VSimplified
gate
charge
waveforms,
(b) Gate
charge
characteristic
. Depending
on
several amperes
amperes
maybeberunning
running
this
state.
GE(th)
GE is still
below
VGE(th)
, nocollector
collectorcurrent
current flows
flows during
maintained
at V
below
VGE(th)
, no
during this
thisperiod
periodand
andVCE
VCEis is
maintained
atCCVlevel.
CC level.
[5, p. 2]
2/8
2/8
2007-10-31 – Rev00
2007-10-31 – Rev00
© by SEMIKRON
© by SEMIKRON
27
4
Matlab Simulation
Continuous
Ideal Switch
powergui
current ia
current ib
g
current ic
C
C
g
C
g
Constant4
+
- v
0
IGBT/S3/Sb
E
Vdc
Scope5
Scope
IGBT/S5/Sc
E
IGBT/S1/Sa
E
Voltage Measurement
+ i
-
Voltage Measurement1
+
- v
Current Measurement
A
+ i
-
B
???
m
Current Measurement1
E
g
Permanent Magnet
Synchronous Machine
C
C
g
IGBT/S6/Sb'
IGBT/S2/Sc'
E
IGBT/S4/Sa'
Voltage Measurement2
E
+
- v
C
g
C
Scope2
current ib
Scope1
Scope4
id_e_synchronous
iq_e_synchronous
desired id_e
ramptime
vd_e_synchronous
Scope3
Constant2
vq_e_synchronous
Product
Rotor speed (rpm)
Gain3
Lq
60/(2*pi)
i_Sdref
PI
desired iq_e
Constant
abc to dq transformation
ramptime
Discrete
PI Controller
S2
vq_e
ramptime
Ts
Repeating
Sequence
Constant6
S3
ramptime
theta_e
S4
S5
fcn
S1
id
iq
S3
S3
fcn
<Rotor angle thetam (rad)>
ic
theta_e
S4
S4
S5
ib
S2
S2
S6
Ts
S5
S6
S6
Gain1
sector
Vdc
t1
P/2
<Rotor angle thetam (rad)>
Space Vector Generator and Gate Drive
Vdc
Constant5
i_Sqref
ia
S1
S1
vd_e
Product1
Constant3
PI
Ld
Gain9
Gain2
Constant1
Discrete
PI Controller1
P/2
<Rotor speed wm (rad/s)>
psim
Figure 23: Matlab simulation of space vector pwm
As seen in figure 23 a Matlab simulation has been created to simulate the
dynamic behaviour of a PMSM. Until now the results from the situation are not
as expected. The simulation needs to be studied in more detail to see where the
problem is. The code for the space vector PWM and Clark-Park transformation
can be found in appendix A.4.
6
5
5
D
4
28
3
Schematic
5.1
Power Supply Module
Voltage Supplies
+12V
D-SUB Connector Power
HEATSINK
F1
D17
1
+12V
J1-2
+12V
J1-3
J2-1
C
1
J1-5
J2-3
IN
OUT
3
GND
C1
330nF
2
J1-6
J2-4
+5V
U1
J1-4
J2-2
2
HEAT1
J1-1
LM7805
C2
100nF
J1-7
J2-5
J2-6
+12V_ISOLATED
+12V
J1-8
J1-9
AEE03A18-L
J2-7
2
J2-8
C34
100µF
+
1
+IN
+OUT
-IN
-OUT
3
C35
10µF
4
+
DCDC1
J2-9
GND_ISOLATED
J2-10
J2-11
B
+12V_ISOLATED
J2-12
+5V_ISOLATED
U6
J2-13
1
J2-14
C13
330nF
J2-15
IN
OUT
3
GND
2
LM7805
C14
100nF
J2-16
J2-17
+5V_ISOLATED
J2-18
J2-19
Figure 24: Power supply schematic
J2-20
A
In figure 24 the schematic of the power supply can be seen. The 12VDRAWN:
power
<Drawn By>
from the car’s system is converted to 5V by a linear voltage regulator to supply
CHECKED:
the low voltage parts. To supply the galvanically isolated parts on the high voltage
<Checked By>
QUALITY
CONTROL:
side a DC to DC inverter is used. Because some parts on the high voltage
side
<QC By>
require 5V as well another linear voltage regulator is used to supply 5V. RELEASED:
<Released By>
D
<
D
<
D
<
D
<
GTC-16026
3
PE6/INT.6/AIN.0
10nF 11 PE0/WR
C5
GND
C6
AREF
SCL
SCL2
2
1
SDA TEMP_OS2
1k2R R54
NC
NC
GND
GND2
GND
GND2
GND
GND2
13
12
11
39
40
C25
C21
C24
C22
PF4/TCK_TEENSY
C20
2
1
D49
1k2R
R102
R2D_A0
10nF
120R
R101
C52
2
1
10nF
2
C53
10nF
D52
1k2R
R108
1
R2D_LOT
D
PF6/TDO_TEENSY
PF7/TDI_TEENSY
R2D_RESET
D171
2
D53
1k2R
R110
1
PF5/TMS_TEENSY
R2D_A1
C51
1R104
C17
C50
C16
C15
+5V
RST
43
42
41
2
1k2R
SPI_MISO_C2000
PF3_TEENSY
R2D_WR/FSYNC
120R
R109
R2D_SAMPLE
120R
R107
C
SPI_CLK_C2000
1
1
93
2 92
NC
R58 120R
PE0_TEENSY
NC
PE1_TEENSY
NC
OVER_CURRENT_FAULT_RESET
NC
NC
NC
NC
NC
SDA_ISOLATED
NC
SCL_ISOLATED
NC
TEMP_OS_ISOLATED
NC
NC
NC
NC
NC
NC
NC
NC
TEMP_OS_ISOLATED
NC
NC
EPWM2A
EPWM3A
EPWM1A
NC
+5V
CURRENT_SENS3_OUT
NC
NC
NC
NC
GND
NC
CURRENT_SENS1_OUT
NC
+5V
NC
INVERTERMODULEMOSFET
2
1
J1-3
J1-4
J1-5
J1-6
R119 1k2R
2
R120 1k2R
C59
J1-7
J1-8
10nF
C60
J1-9
10nF
J1-10
R2D_WR/FSYNC
EPWM3B
EPWM2B
EPWM3A
R2D_SAMPLE
EPWM2A
EPWM1A
ADCINA6
GPIO28/SCIRXDA/SDAA/TZ2
GPIO29/SCITXDA/SCLA/TZ3
GPIO34/COMP2OUT/RSVD/RSVD
ADCINA4
ADCINA2
ADCINB2
+3.3V
86
+5V
85
84
83
120R
R115
10nF
10nF
10nF C35
J6-10
2
NC
J6-9
2
NC
RESET
GPIO16^32/SPISIMOA^SDAA/RSVD^EPWMSYNCI
GPIO17^33/SPISOMIA^SCLA/RSVD^EPWMSYNCO
GPIO6/EPWM4A/EPWMSYNCI/EPWMSYNCO
GPIO7/EPWM4B/SCIRXDA/RSVD
J3-1
GND
J3-2
GND
J3-3
+3.3V
87
GPIO12/TZ1/SCITXDA/RSVD
ADCINB4
89
88
R70 1k2R
C34
1
1 R72 1k2R 2
GPIO18/SPICLK/SCITXDA
GND
GPIO19/SPISTEA/SCIRXDA/ECAP1
ADCINB6
80
1k2R R53
79
2
SDA
J2-3
C44 2
SCL
J2-4
C45 10nF
J2-5
J2-7
2
J2-9
C56
J2-10
10nF
C57
1
2
120R R93
2
R58 120R
R118 1k2R
R60 1k2R
2
1
1
120RR89
120R R97 2
10nF
2
1
R2D_A1
R55 120R
1
SPI_MOSI_C2000
R119 1k2R
C59
SPI_MISO_C2000
2
R2D_LOT
R120 1k2R
10nF
1
R2D_RESET
C60
120RR95
10nF
10nF
C2000LAUNCHPAD
2
120R R125
2
2
120R R124
1
CURRENT_SENS1_OUT
1
CURRENT_SENS2_OUT
1
CURRENT_SENS3_OUT
120R R123
77
A
76
Figure 26: C2000 launchpad schematic
74
COMPANY:
<Company Name>
73
72
SDA_ISOLATED
71
SCL_ISOLATED
70
TEMP_OS_ISOLATED
TITLE:
DRAWN:
QUALITY CONTROL:
65
64
63
62
61
<Checked Date>
CODE:
SIZE:
DRAWING NO:
REV:
EPWM3A
67
66
<Drawn Date>
DATED:
<Checked By>
68
<Title>
DATED:
<Drawn By>
CHECKED:
69
<QC By>
CURRENT_SENS3_OUT
10
1
78
75
2
R56 1k2R
2
1
1
120R R91
2
2
1
2
2
R57 120R
1
R2D_A0
1
R59 120R
1
1
D28
D30
1
120R R87 2
2
1
C46 SPI_CLK_C2000
1k2R R98
2
1
1k2R R116
C4710nF
2
1
J2-8
2
1
FET1
TEMP_OS
1k2R R96
10nF
J2-6
1k2R R92
1
1k2R R94
+3
2
1
1k2R R121
2
1
81
J2-2
10nF
1k2R
R122
2
1
82
J2-1
R85 47k
+3.3V
1
J1-2
47k R86
10nF
R56 1k2R
C30
1
C29
R55 120R
2
1
91
90
10nF
D57
R60 1k2R
2
2
1
2
NC
R57
120R
PC0_TEENSY
EPWM1B
C36
J1-1
D28
D30
1
C2000 Launch
D29
NC
942
D58
95
J2-10
FET2
96
R59 120R
B
PF7/TDI_TEENSY
J2-9
1
1
TEMP_OS
PC1_TEENSY
NC
NC
FET1
+3.3V
J2-8
B1
D46
SCL
JTAG_NSRST
J2-7
D44
PC2_TEENSY
NC
PF5/TMS_TEENSY
J2-6
D45
SDA
J2-5
D43
PC3_TEENSY
97
J2-4
D42
PC4_TEENSY
C2000 Launchpad
98
+5V
PF6/TDO_TEENSY
J2-3
NC
NC
99
PF4/TCK_TEENSY
J5-10
J2-10
100
NC
J2-9
PC5_TEENSY
JTAG Connector
J2-2
ADCINB7
PC6_TEENSY
NC
101
J5-9
NC
+3.3V
PF7/TDI_TEENSY
J1-8
10nF
J2-8
102
J6-8
PC7_TEENSY
NC
SCL_ISOLATED
60
PF2_TEENSY
1
R40
1k2R
10nF
J2-1
GPIO17^33/SPISOMIA^SCLA/RSVD^EPWMSYNCO
NC
J2-7
103
ADCINB3
J2-6
NC
D34
104
10nF C37
GND
JTAG_NSRST
D38
C39
NC
PF5/TMS_TEENSY
D39
D35
105
10nFC38
NC
J2-4
106
J6-7
J2-3
J2-5
NC
NC
+5V
PF6/TDO_TEENSY
J6-6
NC
D47
D36
J6-5
+5V
NC
J2-2
NC
59
PF1_TEENSY
D23
120R R37
J1-9
D56
D37
107
PF4/TCK_TEENSY
NC
58
1
CANMODULE
108
GPIO4/EPWM3A/RSVD/RSVD
OVERCURRENT3_FAULT
109
GPIO5/EPWM3B/RSVD/ECAP1
OVERCURRENT2_FAULT
NC
J2-1
110
GPIO3/EPWM2B/RSVD/COMP2OUT
+12V_ISOLATED
JTAG Connector
OVERCURRENT1_FAULT
SDA_ISOLATED
57
120RR45
CANH
GPIO16^32/SPISIMOA^SDAA/RSVD^EPWMSYNCI
NC
111
ADCINB1
8
9
CURRENT_SENS2_OUT
10
NC
56
2
1
PF0_TEENSY
2
1
J1-7
CANL
ADCINA0
+5V_ISOLATED
112
1
2
3
4
5
6
7
8
9
10
GND
CANL
GND
NC
NC
NC
CANH
NC
NC
GND
ADCINA1
NC
J1-9
6
7
CURRENT_SENS1_OUT
CANH
113
VCC
SPI_CS_TEENSY
SPI_MISO_TEENSY
SPI_MOSI_TEENSY
SPI_CLK_TEENSY
SPI_INT_TEENSY
NC
GND
GND
GND
10nF
CANL
EPWM3B
5
NC
J1-8
NC
55
10nF
1k2R R38
1
R46
1k2R
J1-6
120RR77
1
R79
2 120R 1
120RR71
1
R73
2 120R 1
R75
2 120R 1
R69
2 120R 1
2 120R 1
R117
2 2
1
J1-7
EPWM1B
3
4
NC
20
19
18
17
16
15
14
13
12
11
SPI_SS_TEENSY
SPI_MISO_TEENSY
SPI_MOSI_TEENSY
SPI_CLK_TEENSY
INT.7_TEENSY
J6-1
CANH
NC
54
2
2
J1-6
EPWM2A
114
CANH
2
NC
J6-4
J1-5
EPWM2B
CANMODULE1
115
J6-3
NC
116
J6-2
NC
J1-4
+5V
NC
53
120R R43
D22
2
D18
2
1
1
21
J1-5
117
GPIO2/EPWM2A/RSVD/RSVD
NC
J1-3
EPWM3A
NC
52
1k2RR44
120RR35
22
J1-4
GPIO0/EPWM1A/RSVD/RSVD
NC
NC
51
2
1
JTAG_NSRST
D59
38
1k2R R36
D60
37
37
D27
36
36 10nF
1
118
J5-8
35
2
J1-2
J5-7
34
10nF
35
2
119
J5-6
33
GND
C13
J1-2
J5-5
32
DATE:
1k2R R54
C27
31
APPROVED:
OVERCURRENT1_FAULT
10nF
30
ECO NO:
10nF
29
R84
1
NC
CANL
10nF C28
28
SPI_SS_TEENSY
LTR
1
GND_ISOLATED
J1-1
NC
GPIO1/EPWM1B/RSVD/COMP1OUT
27
2
34
C19
C10
C11
C23
CAN Connector
NC
ADCINA3
26
D26
120
ADCINA7
25
10nF
J1-3
2
24
1
REVISION
RECORD
INT.7_TEENSY
J1-1
GND
23
2
CANL
+5V
22
120R R65
2
NC
J5-4
21
2
1
1
D31
TEENSY++2.0
NC
J5-3
20
1k2R R66
120RR51
32
29
R2D_SAMPLE
GND
J5-2
19
10nF
30
R2D_WR/FSYNC
SPI_CLK_C2000
C61
18
2
2
CAN Connector
C42
17
1k2R R52
2
1
SPI_CLK_TEENSY
1
D25
120R R39
2
2
1
1
10nF
D19
120R R29
1k2R R30
22
1 1
38
2 2
1
10nF
D20
34
PF5/ADC5/TMS
10nF
D48
R2D_A0
1k2R
120RR991k2R R32
R100
2120R R311
2
1
33
39 10nF
2
2
R2D_A1
1
1
D15
10nF
D51
PF6/ADC6/TDO
1k2R
120R
R105
32
R106
2
2
R2D_LOT
1
1
1k2R R34
120RR33
10nF
31
40
2
2
1
1
10nF 1k2R D50
D16
120R
R2D_RESET
R103
PF7/ADC7/TDI
PF4/ADC4/TCK
C43
16
10nF
SPI_MOSI_TEENSY
2
1
In figure 25 and 26 the schematic of the Teensy2++ and the C2000 launchpad
can be seen. The I/Os of the microcontrollers are protected by a zener diode
against voltage transients. Also a resistor directly at the input protects the interior
diodes of the microcontroller against too high currents.
10nF
15
2
C55
J5-1
14
C54
NC
1
CANMODULE
20
1
D17
GND
GND
10
2
CANL EPWM1A
GND
11
NC
NC
12NC
CANH
13NC
NC
GND
PF7/TDI_TEENSY
C30
9
CANMODULE1
1
C29
8
PF3/ADC3
D24
120RR130
D69
30
31
2
120RR49
1
Figure 25: Teensy2++ schematic
PF6/TDO_TEENSY
120RR33
D57
7
2
1
2
6
1k2R R34
R85 47k
5
1k2R R131
1
1k2R R50
RESOLVER2DIGITALMODULE
1
4
40
PF2/ADC2
39
SPI_MOSI_C2000
19
JTAG_NSRST
TEMP_OS_ISOLATED
14
29 10nF
GND
SPI_MISO_C2000
18
R83
SCL_ISOLATED
15
17
PF5/TMS_TEENSY
D29
3
PF4/TCK_TEENSY
SDA_ISOLATED
16
2
SPI_MISO_TEENSY
120RR47
1
2
10nF
17
10nF
28
C18
C7
PF1/ADC1
+12V
C12
C14
10nF
1
NC
TEMP_OS SDA2
INV1
2
2
120R R39
D20
18
TEENSY++2.0
NC
OPTOCOUPLERMODULE
1
1
D16
19
PF3_TEENSY
2
1
27
1
1k2RR48
1
2
D21
33
PF0/ADC0+12V
C9
10nF
1
R40
1k2R
2
10nF
120R R31
2
1
2
2
1
D55 1k2R
R114
9
10
2
D15
2
D19
1
1k2R R32
1k2R
1
SPI_MISO_C2000
2
8
GND2
1
10nF
120R R29
2
1
R41
120R
2
1
R113120R
7
39 10nF
1k2R R30
PF2_TEENSY
1
6
2
D23
120R R37
SPI_CLK_C2000
5
38
+5V2
PF7/ADC7/TDI
NC
10nF
R42
1k2R
PF1_TEENSY
10nF
SCL
SDA
37
20 40
+5V
2
D18
1
1k2R R38
1
2
1
D54 1k2R
R112
4
120RR35
120RR45
2
41
3
2
1
PF0_TEENSY
2
1
R111120R
PC7/A15/IC.3/CLKO
TEMP_OS
36 10nF
1k2R R36
1
R46
1k2R
2
2
26 10nF
C26
PE7/INT.7/AIN.1/UVCON
PD7/T0
25
C23
PD6/T1
24
C48
10
23
C58
2
PB0/SS/PCINT0
22
C33
PB1/PCINT1/SCLK
21
C49
C40
C1
C31
C41
C3
1k2R
2
1 R8
D4
PB2/PDI/PCINT2/MOSI
PD4/ICP1
9
1k2R
2
1 R10
2
PB3/PDO/PCINT3/MISO
10nF 8 PD5/XCK1
10nF
2
PB4/PCINT4/OC.2A
PD3/TXD1/INT3
7
SPI_MOSI_C2000
1
2
TEENSY
O_TEENSY
I_TEENSY
_TEENSY
_TEENSY
D1
PD1/OC2B/SDA/INT1
C8
PF6/ADC6/TDO
2
2
D22
47k R86
10nF 20
OPTO1
PC6/A14/OC.3A
35
1
FET2
19
10nF
1
2
PF4/ADC4/TCK
PF5/ADC5/TMS
PC5/A13/OC.3B
RST
18
2
R7 120R
C25
C26
PF3/ADC3
PC3/A11/T.3
10nF 17 PC4/A12/OC.3C
2
34
C20
PF2/ADC2
PC2/A10
33
C17
16
1
1k2R
2
1 R2
C4
+12V
PF1/ADC1
2
6
C24
PF0/ADC0
15
C10
C21
PC0/A8
GND
C7
13
10nF 14 PC1/A9
C22
12
R9 120R
4
PB7/PCINT7/OC.0A/OC.1C
NC PB6/PCINT6/OC.1B
C19
10nF 11
1
D40
PB5/PCINT5/OC.1A
10nF 5 PD2/RXD1/INT2
10nF
R1 120R
D2
10nF
PE1_TEENSY
C16
9
10
D3
4
1 R4 1k2R 2
2
R6 1k2R
2
C58
10nF 8
2
1
2
PD0/OC0B/SCL/INT0
40
R12 1k2R
+12V
R11120R
39
21
2
2
12
1
1
D5
10nF
PE0_TEENSY
VCC+12V
PE1/RD
38
+12V
R13 120R
R14 1k2R
22
37
D6
2
2
1
1
10nF
PC7_TEENSY
+12V13 PC0/A8
NC
5
36
NC
NC
23
1k2R
R16
120R
R15
PB5/PCINT5/OC.1A
6
R2D1
PD0/OC0B/SCL/INT0
2
2
1
1
10nF 14 PC1/A9
PC6_TEENSY
D7
NC
7
1
NC PB4/PCINT4/OC.2A 24
R18
R42
PD1/OC2B/SDA/INT1
+12V
R17120R
1k2R
R41
8
2
2
15 2
1
1
D8
10nF
NC
PC5_TEENSY
1k2R
PC2/A10
9
25
2
2 120R 1
1
NC
SPI_MISO_TEENSY
PB3/PDO/PCINT3/MISO
PD2/RXD1/INT2 GND
R20 1k2R
10
31
R19120R
R47
1k2RR48
120R
2
2
16
GND
+12V_ISOLATED
1
1
D9
10nF
26 10nF30
2
2 PC4_TEENSY
1
1
D21
SPI_MOSI_TEENSY
11
PC3/A11/T.3
PB2/PDI/PCINT2/MOSI
PD3/TXD1/INT3
GND
+12V_ISOLATED
R22 1k2R
12
29
R49
1k2R R50
R21120R
27
D24
2
2 120R 1
1
NC
+12V_ISOLATED
10nF
2
2
1
SPI_CLK_TEENSY 1
10nF 17 PC4/A12/OC.3C
PC3_TEENSY
D10
PB1/PCINT1/SCLK
13
28
PD4/ICP1
NC
NC
120RR130
14
272 1k2R R1311
R24 1k2R
R23120RSPI_SS_TEENSY
28
2
D25
1
10nF
GND_ISOLATED
PB0/SS/PCINT0
2
2
D11
18
1
1
10nF
PD5/XCK1 15 NC
PC2_TEENSY
26
PC5/A13/OC.3B
1k2R R52
120RR51
NC
GND_ISOLATED
29 10nF25
2
2
1
1
16
D69
INT.7_TEENSY
R26 1k2R
R25120R
PE7/INT.7/AIN.1/UVCON
NC
GND_ISOLATED
PD6/T1
D12
2
2
19
1
1
10nF
17
24
PC1_TEENSY
120R R65
1k2R R66
9PC6/A14/OC.3A
NC
NC
30
2
2
1
1
10nF
D26
NC
OVERCURRENT1_FAULT
18
23
PE6/INT.6/AIN.0
PD7/T0
R28 1k2R
R27120R
+5V
+5V_ISOLATED
2
2
20 10 GND
1
1
19
22
D13
10nF
PC0_TEENSY
31
PC7/A15/IC.3/CLKO
10nF
D31
+5V
+5V_ISOLATED
11
GND
20
21
PE0/WR
NC
+5V
+5V_ISOLATED
D14
10nF
32
AREF
PE1/RD
POWERSUPPLYMODULE
1k2RR44
120R R43
GND
+5V
C1
7
1
+12V
2
+12V
3
+12V
1
3
R3 120R
C33
6
2
POWER1
C18
10nF 5
2
1
D32
10nF
1
R5 120R
1
PD7_LCD_RS
43
C40
4
R641k2R
2
PD5_LCD_EN
C15
+12V
3
1
D33
OVER_CURRENT_FAULT_RESET
4
B18
2
2
C2
R63120R
1
C32
OVERCURRENT2_FAULT
R67120R
1
R62
1k2R
2
10nF
R61 120R
2
1
OVERCURRENT3_FAULT
PB7/PCINT7/OC.0A/OC.1CPB6/PCINT6/OC.1B
3
D41
SDA
PD6_LCD_RW
C13
k2R
R81 120R
2
1
SCL
VCC
GND
2
Teensy2++
1
4
B18
1
VSS
VDD
V0
RS
R/W
EN
DB0
DB1
DB2
DB3
DB4
DB5
DB6
DB7
LCD1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1k2R
2
k2R
Microcontroller
Modules
5
Teensy2++
42
k2R
5.2
6
C4
k2R
29
RELEASED:
<Released By>
DATED:
<QC Date>
<Code> A1<Drawing Number>
<Revision>
DATED:
<Release Date>SCALE: <Scale>
SHEET:
1OF
1
30
Optocoupler Isolation
+5V_ISOLATED
+5V
+5V
SCL 7 SCL
SCL
+
P82B96
7
VO1
+
5
1
2
1
1
-
6
VO2
TEMP_OS
J2-10
B16
VCC
U4-A
2
8
J2-9
VF1
J2-8
2
2
OC
1
U4-B
3
VF2
4
R27
HCPL2630
GND
SCL_ISOLAT
TEMP_OS_ISOLAT
5
GND
RXD,SCL 6
J2-7
1k
1
4
SDA_ISOLAT
390
VF1
2
2
TXD,SCL 5
SDA_ISOLATED
SCL_ISOLATED
2
SDA
3
6
VO2
1
J2-6
RXD,SDA 2
-
7
VO1
390
SCL
2
2
8
VCC
+
VF2
J2-5
1
TXD,SDA 3
1 SDA
SDA
B11
2
8
2
2
TEMP_OS
J2-4
VCC
1k
R1
B15
J2-3
R34
390R
R28
1
1
390R
R25
1
R24
390
R26
1
¨1k5
1k5
R23
J2-2
R22
1
J2-1
7417
4
4
+
GND
HCPL2630
OC
3
TEMP_OS_ISOLATED
7417
Figure 27: Optocoupler module schematic
COMPANY:
5.3
In figure 27 the optocoupler module can be seen. <Drawn
It uses
the<Drawn
P82B96
IC (inteBy>
Date>
grated circuit) and two fast optocouplers to galvanically isolate the high voltage
<Checked By>
<Checked Date>
side from the low voltage side of the I2C bus. This is to protect the low voltage
<QC Date>
<Code>
4
3 <QC By>
side from a5failure on the high voltage
side and to prevent
high voltage 2transients
<Released By>
<Release Date>
<Scale>
coupled into the bus lines.
DRAWN:
DATED:
CHECKED:
DATED:
QUALITY CONTROL:
DATED:
RELEASED:
DATED:
CODE:
SCALE:
D
<Co
TITLE:
5.4
LTR
SIZE:
B <D
1
REVISION RECOR
ECO NO:
CAN Bus Module
T1
10nH
CAN Bus
R1
60R 1
1
2
B1
2
B31
2
C1
330nF
6
Optocoupler Module
CANH
+5V
B30
SPI_MOSI_TEENSY
SPI_CLK_TEENSY
SPI_CS_TEENSY
8
RS
7
CANH
6
CANL
5
VREF
60R 1
R41
2
CANL
MCP2551
C31
100nF
C16
C15
330nF
C17
330nF
330nF
C18
330nF
1
2
A
LED
D16
2
Y2
560R
R66
560R
R65
2
A
K
D15
1
TXD
2
VSS
3
VDD
4
RXD
MCP2515
1
SPI_INT_TEENSY
1
2
3
4
5
6
7
8
9
LED
C
1
B34
VDD
TXCAN
RESET
RXCAN
CS CLKOUT/SOF
SO
TX0RTS
SI
TX1RTS
SCK
TX2RTS
INT
OSC2
RX0BF
OSC1
RX1BF
VSS
D10
SPI_MISO_TEENSY
18
17
16
15
14
13
12
11
10
1
1
22pF
K
2
10k
R62
20MHz
C32
C33
22pF
GND
B
Figure 28: CAN bus module schematic
J1-1
J1-2
SPI_CS_TEENSY
J1-3
SPI_MISO_TEENSY
J2-1
J2-2
CANL
In figure 28 the CAN bus module can be seen. It uses the MCP2515 as the SPI
<Company Name>
Can interface and the MCP2551 bus driver. A common mode choke is considered
in case of interference that leads to a bad bus. The choke is not placed on the <Title>
J1-4
SPI_MOSI_TEENSY
J1-5
SPI_CLK_TEENSY
J1-6
SPI_INT_TEENSY
COMPANY:
J2-3
J2-4
TITLE:
J2-5
J2-6
J1-7
A
J2-7
DRAWN:
CANH
DATED:
<Drawn By>
<Drawn Date>
J1-8
J2-8
J1-9
CHECKED:
DATED:
<Checked By>
<Checked Date>
CODE:
SIZE:
DRAWING NO:
J2-9
J1-10
QUALITY CONTROL:
<QC By>
J2-10
RELEASED:
<Released By>
DATED:
<QC Date>
<Code>
B <Drawing Num
DATED:
<Release Date>SCALE: <Scale>
SHEET:
2
7417
1
D
J1-8
CURRENT_SENS2_OUT
J1-9
CURRENT_SENS1_OUT
J1-12
EPWM1B
J1-13
EPWM2A
J1-14
EPWM2B
J1-15
4
+
5
GND
R31
PC0_TEENSY
PC1_TEENSY
PC2_TEENSY
PC3_TEENSY
PC4_TEENSY
PC5_TEENSY
PC6_TEENSY
PC7_TEENSY
11
10
9
8
7
6
5
4
PE0_TEENSY
PE1_TEENSY
12
13
B0
B1
B2
B3
B4
B5
B6
B7
VDD
3S2
10
3R
9
3Q
SN54LS279AN
HCPL2630
31
U5-A
1
OC
2
OVERCURRENT1_FAULT
14
U5-B
74ALS05A
3
RFB
I_OUT1
I_OUT2
OC
4
16
board yet. If the CAN bus works without the choke it is better to avoid using it
because it creates high voltage transients at the driver’s bus input.
74ALS05A
1
2
OVERCURRENT_VOLTAGE_REF
CS
WR
3
GND
EPWM3A
J1-16
6
VO2
U8
EPWM1A
J1-11
2
1
U4-A
EPWM3B
J1-10
2
OC
OVER_CURRENT_FAULT_RESET
2S
7
2Q
8
GND
VO1
-
3
390R
2
R26
V
-
J1-7
VF2
2
OVERCURRENT_VOLTAGE_REF
J1-6
VREF
AD7524N
15
100µF
C15
J1-17
J1-18
5.5
J1-19
Inverter Module
J1-20
Temperature Sensor
Current Measurement
B8
1
1
10k
R24
EPWM1A
2 2
CURRENT_SENS1_OUT
2P-2.54
5
6
7
SE95
8
2
C
3
4
1
1
8
VCC
7
A0
6
A1
5
A2
JUMPER4
10k
B12
1
SDA
2
SCL
3
OS
4
GND
SDA_ISOLATED
SCL_ISOLATED
TEMP_OS_ISOLATED
2
+
C2
R33
J2-2
2
C1
100nF
R22
9
1
4M7
F1
10
VDD1
IIN
VDD2
IIN
R1
IIN
R2
IIN
GND2
IIN
GND3
IOUT
OUT
IOUT
NC
IOUT
TRST
IOUT
GND1
IOUT
20
19
SDA_ISOLATED
SCL_ISOLATED
CURRENT_SENS1
TEMP_OS_ISOLATED
18
17
16
15
14
13
J4-1
12
11
SI8503-C-IS
+12V_ISOLATED
D1
B2
1
EPWM2A
1
2
+
1k
4
10
B3
1
R50
1
4R7
J2-1
Figure 29: Mosfet inverter module schematic
J3-1
J6-2
J6-3
J6-4
J6-5
J6-6
J6-7
In figure 29 the schematic of the inverter module can be seen. It includes the
ADC for measuring the battery’s voltage, the isolated bootstrap driver circuit, the
temperature sensor for the pair of mosfets, the current measurement and the two
mosfet switches.
J3-2
J3-3
J3-4
TEMP_OS_ISOLATED
SCL_ISOLATED
SDA_ISOLATED
OVER_CURRENT_FAULT_RESET
J6-10
PE1_TEENSY
J6-12
J6-13
J6-14
J6-15
J6-16
J6-17
J6-18
J6-19
J6-20
J3-7
J3-9
J6-9
J6-11
J3-6
J3-8
J6-8
A
J3-5
5.6
Resolver to Digital Converter Module
PE0_TEENSY
PC0_TEENSY
PC1_TEENSY
J3-10
J3-11
J3-12
J3-13
J3-14
In figure 30 the schematic for the resolver to digital converter module can be
seen. The motor’s resolver is connected at the terminals S1-S4 and R1-R2. R1
and R2 supply the excitation voltage. S1-S4 receive the angle information of the
PC2_TEENSY
PC3_TEENSY
PC4_TEENSY
PC5_TEENSY
PC6_TEENSY
PC7_TEENSY
J3-15
J3-16
J3-17
J3-18
J3-19
J3-20
7417
R14
7417
2
HCPL2630
R7
OC
C4
U1-E
+
2 1
2
3
390R
1
11
74LS08
6
D2
U9-B
B
J6-1
OC
8
R13
10k
5
390R
1
FAN7382
2
GND
CURRENT_SENS1
8
VB
7
HO
6
VS
5
LO
U1-D
U2
1
VCC
2
HIN
3
LIN
4
COM
C3
150nF
D6
3
1
R1
6
VO2
5
+
9
1
R2
2
4
7
VO1
2
-
3
U9-A
74LS08
EPWM2B
+C11
2
2
OVERCURRENT1_FAULT
C12
7417
OC
U1-C
5
SDA_ISOLATED
R3
VCC
+
1
8
VF2
4
6
1
VF1
1
1k
3
OC
U1-B
7417
B10
1
R30
46k9
390R
EPWM1B
SCL_ISOLATED
2
2
2
2
3
MCP3021A5T
AIN
SDA
VSS
390R
1
R4
4
5
SCL
ADC1
VDD
1
HV Measuring
1
2
10k
+5V_ISOLATED
EPWM1A
R8
4R7
5R
Isolation
2
R49
+
R9
R29
C9
2
+
OVERCURRENT1_FAULT
996k5
1
Bootstrap Driver
C10
2 4
C29
J3-9
J3-8
J3-7
J3-6
J3-5
J3-4
120pF
+12V
R42
S3
R47
2
1
1
15.4k
S1
S2
S4
2
2.2k
C
Q1
2
B
R45
2
R64
2
1
R1
2
4R7
2
1
R48
3R3
1
E
1
R2D_SAMPLE
Q2
B
R2D_WR/FSYNC
C
R2D_RESET
R49
2
R2D_RESET
1
2.2k
R2D_LOT
R2D_A1
120pF
R2D_A0
+12V
AD2S1210
13
14
15
16
17
18
19
20
21
22
23
24
2
R52
2
1
15.4k
R57
1
2.2k
C
Q3
2
B
C2
4.7µF
2
+
R60
C27
10nF
E
4R7
SPI_CLK_C2000
R56
SPI_MOSI_C2000
SPI_MISO_C2000
R44
2
10k
R2D_A1
3R3
1
C26
22pF
1
1
Y1
R2D_SAMPLE
R43
22k
36
35
34
33
32
31
30
29
28
27
26
25
C30
R2D_WR/FSYNC
2
A1
DOS
LOT
RESET
DIR
NM
B
A
DB0
DB1
DB2
DB3
R2D_LOT
48
47
46
45
44
43
42
41
40
39
38
37
REFOUT
REFBYP
COS
COSLO
AVDD
SINLO
SIN
AGND
EXC
EXC
A0
RES0
4.7µF
+
RES1
CS
RD
WR/FSYNC
DGND
DVDD
CLKIN
XTALOUT
SOE
SAMPLE
DB15/SDO
DB14/SDI
DB13/SCLK
DB12
DB11
DB10
DB9
VDRIVE
DGND
DB8
DB7
DB6
DB5
DB4
1
B36
1
C1
B33-A
TS992
B32
R51
2
+IN1
R2D_A0
1
2
3
4
5
6
7
8
9
10
11
12
D12
1
10k
2
10k
R63
1
B37
C19
2
C20
10µF
+5V
SPI_MOSI_C2000
-IN1
OUT1
10nF
2
22pF
8.192MHz
C25
2
1
10k
3
C24
10nF
+5V
1
C21
10nF
R50
1
+
D11
2
C22
4.7µF
+5V
SPI_MISO_C2000
4R7
+12V
E
R46
3R3
+5V
C3
680nF
SPI_CLK_C2000
6
1
D13
1
R55
2
-IN2
10k
R2
B33-B
2
1
10k
2
1
1
22k
R58
2
4R7
TS992
R54
3R3
2
R53
R61
+IN2
+5V
7
D14
OUT2
5
E
1
J3-3
J3-2
32
Q4
B
C
2
R59
1
2.2k
Figure 30: Resolver to digital converter module schematic
COMPANY:
rotor. The output voltages of the resolver that are interpreted by the converter are
expressed as sine and cosine functions of the rotor angle θr :
<Drawn By>
<Drawn Date>
DRAWN:
DATED:
CHECKED:
DATED:
<Checked By>
Vd = V0 sin θr sin ωrs t
QUALITY CONTROL:
<QC By>
RELEASED:
Vq = V0 cos θr sin ωrs t
<
TITLE:
<Released By>
<Checked Date>
DATED:
<QC Date>
CODE:
SIZ
<Code> A
DATED:
<Release Date>SCALE: <Scale>
33
6
Inverter Parts
Power Supply
Module
Teensy2++
In figure 31 the assembled mosfet inverter can be seen. It is separated into
8 modules with the purpose to make failure search easier. Being a prototype if
something goes bad only the damaged module has to be replaced or redesigned.
LCD
CAN Bus
Module
C2000
Launchpad
Optocoupler
Module
Inverter Module Mosfet
Resolver to
Digital Converter
Module
Figure 31: Modules of the mosfet inverter
6.1
Switching Devices
The choice of the right switch depends on the voltage of the intermediate circuit, the maximum current, the switching frequency and the ambient temperature.
Lower switching frequencies are generally better because the on an off-state is
longer compared to the switching time in which Rdson creates heat. What switching frequencies are permissible for the switches and what frequencies are suitable
for the motor to be controlled has to be tested. For now a maximum switching frequency of 4kHz is assumed.
34
6.1.1
MOSFET Driver
In order to design a good driver we have to look at some requirements:
To achieve a good efficiency with the inverter it is important to keep the switching
losses as small as possible. Most of the losses are created during the on and
off switching of the device. During the transition of these two states the switch
provides a greater resistance and more power is turned into heat. This is due
to the fact that during a transition the mosfet is not ideally driven in saturation.
The n-doped layer is not fully enhanced. A quick transition means less losses.
This quick transition can be achieved by reducing the time it takes to charge and
discharge the gate capacitances of the switch. The charging and discharging time
can be reduced by providing a high enough current.
The mosfets are switching a high voltage power source. The hard switching leads
to electromagnetic interference that can be coupled back into the system and
effect the electronics. Thus it is necessary to galvanically isolate the switches
from the low voltage electronics.
Because each mosfet pair is connected in series a driver has to be implemented
which can drive each mosfet separately without the risk of accidentally switching
on the upper switch. This could be caused by the flyback voltage generated by
the connected motor’s inductance. The flyback voltage can create a negative
potential which can lead the mosfet to switch on again.
A bootstrap driver was chosen because of its simplicity and small cost. It has the
disadvantage that the on-time duty cycle is limited by the requirement to refresh
the charge in the bootstrap capacitor.
6.1.1.1
Bootstrap Driver Calculation
In figure 32 you can see the schematic of the driver circuit. The driver ic used
is FAN7382. The choice for this driver is based on recommendations according
to the application note AN-6076 from Fairchild Semiconductor. The mosfet to be
driven is IXFH-FK170N10P.
SE95
8
2
R22
9
1
4M7
F1
10
NC
IOUT
TRST
IOUT
GND1
IOUT
13
J4-1
12
35
11
SI8503-C-IS
+12V_ISOLATED
ATED
_CURRENT_FAULT_RESET
TEENSY
TEENSY
TEENSY
TEENSY
TEENSY
TEENSY
TEENSY
TEENSY
TEENSY
TEENSY
D1
EPWM2A
1
10k
Ilkgs
Ilkcap
Iqbs
Ilkdiode
Ilk
ton
7417
9
1
R50
EPWM2B
Value
12V
0.7V
10V
198nC
3nC
Description
Supply voltage of gate driver
Bootstrap diode forward voltage drop
minimum gate-source saturation voltage
Total gate charge
Charge required by the internal level shifter, which
J3-1
is set to 3 nC for all HV gate drivers
J3-2
100nA Switch gate-source leakage current
J3-3
0nA
Bootstrap capacitor leakage current
J3-4
120µA Bootstrap circuit quiescent current
J3-5
10nA Bootstrap diode leakage current J3-6
50µA Bootstrap circuit leakage current J3-7
J3-8
25µs High-side switch on time
Table 4: Bootstrap driver parameters J3-9
J3-10
J3-11
J3-12
The total charge that needs to be supplied by the bootstrapJ3-13
capacitor is calculated by the following equation.
J3-14
J3-15
J3-16
J3-17
Qtotal = Qg + (Ilkcap + Ilkgs + Iqbs + Ilk + Ilkdiode ) · ton + QJ3-18
ls = 205.3nC
J3-19
J3-20
The voltage over the capacitor is determined by the supplied
driver voltage,
the voltage drop over the diode and the minimum voltage to drive the switch into
R14
7417
B3
1
The parameters involved in the calculations are:
Parameter name
Vdd
Vf
Vgsmin
Qg
Qls
OC
8
R7
4R7
OC
2
10
1
C4
U1-E
74LS08
390R
+
11
U9-B
HCPL2630
+
U1-D
D6
C3
150nF
R13
FAN7382
D2
5
390R
1
6
10k
R1
1k
4
CURRENT_SENS1
8
VB
7
HO
6
VS
5
LO
2
7417
OC
U1-C
GND
5
TED
U2
1
VCC
2
HIN
3
LIN
4
COM
Figure 32: Mosfet driver schematic
J2-1
SDA_ISOLATED
ISOLATED
+C11
7
VO1
6
VO2
EPWM1B
C12
3
74LS08
5
+
1k
1
4
U9-A
1
-
3
2
8
VCC
+
1
OVERCURRENT1_FAULT
6
R3
2
VF1
1
R30
46k9
390R
1
2
AIN
SDA
1
2
4
MCP3021A5T
R4
B10
2
2
VF2
3
1
2
R2
7417
3
U1-B
OC
390R
4
Measuring
1
2
EPWM1A
B2
1
2
+5V_ISOLATED
R8
4R7
5R
Isolation
2
R49
+
R9
R29
C9
2
+
OVERCURRENT1_FAULT
996k5
1
Bootstrap Driver
C10
36
saturation. The voltage drop over the resistor is not relevant because it only limits
the time in which the capacitor can be charged.
∆Vboot = Vdd − Vf − Vgsmin = 1.3V
The value of the bootstrap capacitor then is:
Cboot =
Qtotal
= 157.9nF.
∆Vboot
As the bootstrap resistor value 8Ω is chosen. The resistor limits the charging
current of the capacitor to protect it. The RC constant for charging the capacitor
is:
τ = 8Ω · 157.9nF = 1.26µs
This is the minimal time that has to be ensured to exist between the PWM
pulses to allow the capacitor to charge up. With a maximum frequency of 4000Hz
even with a duty cycle of 99% there would be enough time to charge the capacitor.
1
1
· (1 − D) =
· (1 − 0.99) = 2.5µs
f
4kHz
6.1.2
IGBT Driver
The igbt module to be driven is 6MBI 100S-120. It combines a full 3 phase bridge
in one package. The igbt gate charges are nearly equal to that of the mosfet
driver. Here another approach is taken from the application note AN-7004.
The following parameters are considered:
• Qg = 600nC - The dynamic gate charge of the igbt.
• Vgon = 12V - The gate voltage to drive the igbt in saturation.
• Vgof f = 0V - The minimum voltage at which the igbt is shut off.
• Rg = 12Ω - The gate resistance.
• fs = 4kHz - The switching frequency.
• VCE = 400V - The collector-emitter voltage.
• Pdriver - Gate driver output power.
• Igpeak - Gate driver peak current.
The dynamic gate charge can be read from the igbt datasheet seen in figure
33 (a). At VCE = 400V we can read around 200nC. To be on the safe side we can
go a bit higher on the curve. 600nC is chosen. In figure 33 (b) at Vgon = 12V we
can see that the igbt is in saturation.
37
(a)
(b)
Figure 33: (a)Igbt dynamic gate charge, (b VGE vs. current)
The energy needed to be transferred to the gate is:
E = Qg · (Vgon − Vgof f ) = 2nW h.
The driver output power needs to be:
Pdriver = E · fs = 0.029W.
The average gate current is:
Ig = Qg · fs = 2.4mA.
The peak gate current is:
Igpeak =
Vgon − Vgof f
Rg
As an integrated solution the driver module SKYPER 32 R from Semikron
could be used to drive a pair of igbts.
38
6.2
Protection
The inverter has to be protected to ensure the safety of the people working with it
and to protect the device and the motor attached to it. In any case, situations for
which the inverter and the motor are not designed have to be avoided.
Variables that need to be monitored are:
• DC battery voltage. If the DC voltage is to high it can cause the isolation of
the switching devices to break down and destroy them.
• Switch and motor temperature. If the switches experience high temperatures due to inappropriate switching or too high ambient temperature they
could get damaged.
• Current. A short circuit can cause the temperature in the switches and the
motor to rise rapidly, to fast for the temperature control to respond.
• Motor speed. High motor speed above the maximum rating or for longer periods can heat the motor and can also present a mechanical risk to people.
The battery voltage is measured with a 10 bit ADC (Analogue to digital inverter). The ADC is connected to the isolated I2C bus (Inter-integrated circuit).
Overvoltage protection exists in the form of a RDC snubber (directional snubber)
The temperature of the switches is measured by placing the heatsink of a pair of
mosfets on a SOIC (small-outline integrated circuit) 13 bit temperature sensor.
The motor temperature is measured with 3 analogue sensors which are measured by the C2000 launchpad’s ADCs.
A fuse is placed at the positive battery terminal.
6.2.1
Short Circuit Protection
The current is measured with 3 AC sensors with an analog 5V output.
Short circuit protection is implemented with logic gates. In the case of a short circuit the logic will latch and will not return into operational mode without a manual
failure reset. The logic circuit is shown in figure 34. The comparator to the left
switches it output to a logic one if the current voltage signals rises higher than
the reference voltage at the negative input. This sets the latch (flip-flop) to a logic
one. The signal is then inverted and fed to an AND gate. Thus blocking any input
from the PWM input to the AND gate and preventing the mosfet to be switched.
Figure 34: Short circuit protection logic circuit
B1
39
B11
2P-2.54
5
6
1
RCD Snubber
CURRENT_SENS2_OUT
SE95
7
8
R23
2
9
1
4M7
10
R1
IIN
R2
IIN
GND2
IIN
GND3
IOUT
OUT
IOUT
NC
IOUT
TRST
IOUT
GND1
IOUT
20
1
1
19
18
C16
100nF
CURRENT_SENS2
C17
2
+
47k
IIN
R41
VDD2
EPWM3A
3
4.7µF
17
4
2 2
2 2
4
JUMPER3
1
6.2.2
SE95
R39
10k
8
VCC
7
A0
6
A1
5
A2
3
IIN
16
15
14
CURRENT_SENS3_OUT
13
5
2P-2.54
EPWM2A
VDD1
JUMPER2
47k
R25
+
1
10k
R36
1
SDA
2
SCL
3
OS
4
GND
SDA_ISOLATED
SCL_ISOLATED
TEMP_OS_ISOLATED
1
8
VCC
7
A0
6
A1
5
A2
R40
10k
2
B25
B21
1
SDA
2
SCL
3
OS
4
GND
C14
2
2
C13
100nF
2
6
1
1
1
7
J4-2
8
12
2
11
R34
9
1
In figure 35 the RCD snubber circuit can be seen.
To design it the following parameters are considered:
4M7
10
VDD1
IIN
VDD2
IIN
R1
IIN
R2
IIN
GND2
IIN
GND3
IOUT
OUT
IOUT
NC
IOUT
TRST
IOUT
GND1
IOUT
20
19
18
CURRENT_SENS3
17
16
15
14
13
J4-3
12
11
J8-1
J7-2
J8-2
J7-3
J8-3
J7-4
J8-4
J7-5
J8-5
J7-6
J8-6
J7-7
J8-7
EPWM1A
J7-8
J8-8
EPWM2A
J7-9
J8-9
J7-10
J8-10
J7-11
J8-11
2
50R
R46
D3
B13
1
C18
15nF
4R7
2
R52
1
OVERCURRENT3_FAULT
1
R15
C19
15nF
B
JUMPER1
1
2
2P-5.08
Figure 35: RCD snubber circuit
Cs = 2 · 2 · (Coss + Cmount ) = 11.36nF
J7-1
10k
1
74LS08
2
EPWM3B
R51
2
CURRENT_SENS3
FAN7382
C8
D8
U10-B
+
D7
1
2
D10
1k
5
6
8
VB
7
HO
6
VS
5
LO
10k
GND
1
10k
U7
1
VCC
2
HIN
3
LIN
4
COM
HCPL2630
2
R48
4
5
+
2
U1-A
2 4
7
VO1
6
VO2
R20
7417
B5
1
OC
R5
4R7
OVERCURRENT2_FAULT
-
VF2
390R
2
VCC
+
B7
1
4R7
1
1
2
3
1
74LS08
HCPL2630
1k
R19
R21
FAN7382
C6
3
U10-A
74LS08
R18
2 1
8
2
1
VF1
12
390R
1
B14
1
2
2
7417
OC
D9
U1-F
CURRENT_SENS2
8
VB
7
HO
6
VS
5
LO
R16
R17
10k
R47
2
13
U10-D
+
2
5R
1
2
11
C5
150nF
U3
1
VCC
2
HIN
3
LIN
4
COM
1k
12
5
EPWM3A
D5
GND
R10
5R
1
1k
R12
2
VF2
6
VO2
13
+
U10-C
10
8
1
-
7
VO1
+5V_ISOLATED
C
+12V_ISOLATED
• Vpeak = 100V - The estimated flyback voltage is
around 2 times the battery’s voltage.
• Ipeak = 20A - maximum estimated bleeding resistor current
• Coss = 2340pF - IXFH-FK170N10P
• Cmount = 500pF - IXFH-FK170N10P
• fs = 4kHz - maximum switching frequency
A safe value for the snubber capacitance is 2 times
the total gate capacitance. Because there are 2 active
switching paths at all times this value can be doubled.
9
74LS08
R11
VCC
+
B4
1
4R7
8
2
VF1
B6
R6
SI8503-C-IS
C7
150nF
2
D4
+12V_ISOLATED
1
+5V_ISOLATED
OVERCURRENT3_FAULT
OVERCURRENT2_FAULT
SI8503-C-IS
J5-1
CURRENT_SENS1_OUT
A standard value for Cs of 15nF is used.
Energy to be dissipated in the capacitor is:
J5-2
J5-3
J5-4
J5-5
CURRENT_SENS3_OUT
ERs
TEMP_OS_ISOLATED
J8-12
J7-13
J8-13
2
Cs · Vpeak
= 15.78nW h.
=
2
J5-7
EPWM3A
J7-12
J5-6
J5-8
J5-9
J5-10
SCL_ISOLATED
J5-11
SDA_ISOLATED
J5-12
A
The average power dissipation in Rso at given frequency is:
<Company Name>
J5-13
J7-14
J8-14
J7-15
J8-15
J7-16
J8-16
J7-17
J8-17
J7-18
J8-18
COMPANY:
J5-14
J5-15
J5-16
PRs = Cs ·
J5-17
J5-18
J7-19
J8-19
J7-20
J8-20
TITLE:
2
Vpeak
fs
<Drawn· By>
DRAWN:
CHECKED:
The value for Rs is:
J5-20
QUALITY CONTROL:
<QC By>
RELEASED:
<Released By>
Rs =
<Drawn Date>
DATED:
<Checked By>
J5-19
<Title>
= 0.454W.
DATED:
<Checked Date>
DATED:
<QC Date>
CODE:
SIZE:
DRAWING NO:
REV:
<Code> A1<Drawing Number>
<Revision>
DATED:
<Release Date>SCALE: <Scale>
Vpeak
= 5Ω.
Ipeak
SHEET:
1OF
1
40
6.3
Microcontroller
As the motorcontroller the C2000
Launchpad from Texas Instruments is
used. The microcontroller itself on the
board is Piccolo TMS320F28027. It
was chosen because of its capability to run complex control algorithms
like field oriented control at a low cost.
The board costs around 14 e(May
2013). Ready made libraries are available that provide basic tools for motor control. This makes the software
development comparatively easy and
saves time.
6.4
6.4.1
Figure 36: C2000 Launchpad
Sensors
Temperature Sensors
As a temperature sensor SE95DP from NXP is used.
It communicates via I2C and has a temperature accu- Figure 37: Temperature
racy of ±1°C at 13-bit resolution. 6 of these sensors Sensor SE95DP from
NXP
are placed beneath the IGBT module. 3 of these sensors are used to sense the temperature of the mosfet
modules, one sensor for each pair.
6.4.2
Current Sensors
For the low power version, which uses MOSFET
switches, a shunt resistor will be used to measure the
current to keep the price low. For the high power version, which uses IGBT switches, closed loop Hall effect
sensor will be used to measure the current with a good
linearity over a wide range. One sensor measures the
DC input current from the battery. Two other sensors
measure the AC phase currents ia and ib . ic can be calculated from those two phase currents. To avoid long
measuring wires the sensors are placed directly at the
Figure 38:
Current
Sensor
HAIS-50-P
and HAIS-100-P from
LEM used for current
measurement.
housing, where the wire
41
is being fed through the cable gland. There the measurement contacts are connected closely to the measuring microcontroller. The DC leading cable coming
from the battery pack has a wire size of 16mm2 and a diameter of 10.2mm. The
wire size has to be considered to fit through the sensing hole. A too tight fit might
damage the sensor.
6.4.3
Voltage Sensor
The DC battery voltage is measured using a 10 bit ADC inverter which is connected to the I2C bus. A voltage divider is used to reduce the measured voltage
to 5V maximum.
42
7
7.1
Measurements
Voltage Divider Resistance
Figure 39: Measurement setup
In figure 39 the measurement setup for measuring the true resistor values
for the voltage divider can be seen. This voltage divider is used to measure the
DC voltage of the battery. When the dc voltage is applied current will heat up the
resistors and change their value. To get good measurements the voltage drop and
current are separately measured over a range of voltages. A couple of minutes
were waited to allow to values to settle. In figure 40 and table 5 the results can
be seen. A linear regression gives the values for the resistors.
43
Figure 40: Linear regression of resistance value
V [V ]
V 1[V ](calc.)
V 2[V ]
I[µA]
97.60
93.23
4.37
103.50
89.80
85.78
4.02
95.25
80.00
76.42
3.58
84.90
70.00
66.87
3.13
74.20
59.90
57.21
2.688
63.60
50.00
47.76
2.241
53.00
40.00
38.21
1.793
42.40
30.10
28.75
1.350
31.90
19.98
19.08
0.895
21.10
10.00
9.55
0.448
10.60
5.00
4.78
0.224
5.30
R1 lin.reg.[kΩ] R2 lin.reg.[kΩ]
900.500
4.22
Table 5: Resistance measurement results
44
7.2
Inverter
The generated PWM is filtered using a RC low pass filter with R = 50kΩ and
C = 8nF .
Figure 41: PWM at 4000 Hz
In figure 41 the generated sinusoidal PWM by the microcontroller can be seen.
A lookup table with 12 lookup points is used resulting in the waveform in figure
42.
Figure 42: Sine wave 166 Hz with 12 lookup points
45
Figure 43: PWM at 25000 Hz
In figure 43 the generated sinusoidal PWM by the microcontroller can be seen.
A lookup table with 42 lookup points is used resulting in the waveform in figure
44.
Figure 44: Sine wave 285 Hz with 42 lookup points
46
8
Program
Can Bus
Controller
SPI Bus
Battery
Teensy
(I2C Master)
TI C2000
Launchpad
Galvanic
Isolation
Mosfet Driver with
Overcurrent Protection
Mosfet
Switches
ADC Input
DC Voltage
Sensor x1
Temperature
Sensor x3
us
IB
SP
I2C Communication Bus
Galvanic isolation
AC Current
Sensor x3
Resolver
Signal
AC Motor
Figure 45: System communication overview
In figure 45 you can see a system overview of the inverter. The systems in
blue are the parts that need to be programmed. Until now the CAN bus communication using SPI has been programmed. The I2C communication with the
temperature sensors has been programmed but not successfully tested. Using
the C2000 launchpad the sinusoidal PWM has been programmed to work in a
stationary state.
The control loop has still to be implemented. The resolver information via SPI
has to be programmed. The I2C routine to read the DC voltage has to be programmed.
47
9
Summary and Conclusion
In this thesis the theory of controlling a permanent magnet synchronous motors
has been introduced. The control loop of the inverter was shown and direct torque
control and field oriented control were discussed. A low voltage mosfet inverter
was built. The sinusoidal pulse width modulation was programmed and tested.
Most of the parts of a inverter were discussed. Ways to design and calculate a
driver for a mosfet or igbt were shown. The components used in the project were
mostly taken from the inventory of the university’s laboratory. Thus the cost was
kept lower than 150 e.
As an outline into the future of this project the following points are considered.
The control equations to design a PI controller have to be discussed.
The low voltage mosfet inverter has to be further tested and programmed.
A field oriented control can then be implemented and tested on a low voltage inverter.
The low voltage mosfet inverter was mostly built with through hole components.
To keep parasitic capacitances and inductances as low as possible the final inverter should be designed using small outline components.
48
List of Figures
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Formula Student logos . . . . . . . . . . . . . . . . . . . . . . . . .
Metropolia Motorsport logo . . . . . . . . . . . . . . . . . . . . . .
System Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Taxonomy of AC motors . . . . . . . . . . . . . . . . . . . . . . . .
(a) Continuous Counter EMF of a PMSM, (b) Trapezoidal Counter
EMF of a BLDC . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Typical PMSM structures: (a) surface magnet, (b) inset magnet,
(c) pole shoe rotor, (d) tangentially embedded magnets, (e) radially
embedded magnets (flux-concentration), (f) two magnets per pole
in the V position, (g) a synchronous reluctance motor equipped
with permanent magnets.[8, p. 397] . . . . . . . . . . . . . . . . .
Structure of the Metropolia Motorsport’s motor: interior magnet . .
Simplified equivalent circuit of a 3 Phase AC motor. . . . . . . . . .
Stationary equivalent circuit of a PMSM (one phase) . . . . . . . .
Square wave conversion . . . . . . . . . . . . . . . . . . . . . . . .
Square wave amplitude spectrum . . . . . . . . . . . . . . . . . . .
Modified sine wave conversion . . . . . . . . . . . . . . . . . . . .
Modified sine wave amplitude spectrum . . . . . . . . . . . . . . .
Simulated PWM signal for a pure sine wave . . . . . . . . . . . . .
Simulated PWM signal for a pure sine wave amplitude spectrum .
Stationary frame of the three phases abc with overlayed stationary
α,β frame (a), Space dector diagram (b) . . . . . . . . . . . . . . .
Space vector synthesizing . . . . . . . . . . . . . . . . . . . . . . .
Control loop overview . . . . . . . . . . . . . . . . . . . . . . . . .
Clarke and Park transformation [7, p. 9] . . . . . . . . . . . . . . .
(a) Equivalent MOSFET representation including junction capacitances; and (b) representation of this physical location.[9, p. 83] . .
The IGBT (a) half-cell vertical cross section and equivalent circuit
model[9, p. 102], (b) IGBT capacitances . . . . . . . . . . . . . . .
(a) Simplified gate charge waveforms, (b) Gate charge characteristic [5, p. 2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Matlab simulation of space vector pwm . . . . . . . . . . . . . . . .
Power supply schematic . . . . . . . . . . . . . . . . . . . . . . . .
Teensy2++ schematic . . . . . . . . . . . . . . . . . . . . . . . . .
C2000 launchpad schematic . . . . . . . . . . . . . . . . . . . . .
Optocoupler module schematic . . . . . . . . . . . . . . . . . . . .
CAN bus module schematic . . . . . . . . . . . . . . . . . . . . . .
3
3
4
6
7
8
9
11
12
15
16
16
17
17
18
19
19
21
23
24
25
26
27
28
29
29
30
30
49
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Mosfet inverter module schematic . . . . . . . . .
Resolver to digital converter module schematic . .
Modules of the mosfet inverter . . . . . . . . . . .
Mosfet driver schematic . . . . . . . . . . . . . . .
(a)Igbt dynamic gate charge, (b VGE vs. current) .
Short circuit protection logic circuit . . . . . . . . .
RCD snubber circuit . . . . . . . . . . . . . . . . .
C2000 Launchpad . . . . . . . . . . . . . . . . . .
Temperature Sensor SE95DP from NXP . . . . . .
Current Sensor HAIS-50-P and HAIS-100-P from
current measurement. . . . . . . . . . . . . . . . .
Measurement setup . . . . . . . . . . . . . . . . .
Linear regression of resistance value . . . . . . . .
PWM at 4000 Hz . . . . . . . . . . . . . . . . . . .
Sine wave 166 Hz with 12 lookup points . . . . . .
PWM at 25000 Hz . . . . . . . . . . . . . . . . . .
Sine wave 285 Hz with 42 lookup points . . . . . .
System communication overview . . . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
LEM used for
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
31
32
33
35
37
38
39
40
40
40
42
43
44
44
45
45
46
50
List of Tables
1
2
3
4
5
PMSM parameter overview . . . . .
PMSM dynamic equations[4, p. 154]
Comparison between DTC and FOC
Bootstrap driver parameters . . . . .
Resistance measurement results . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10
13
22
35
43
51
References
[1] Metropolia
Motorsport
Team
History,
2013.
http://
metropolia-motorsport.fi/team/team-history/; accessed 20.05.2013.
[2] Zsolt Borkai.
About FSHungary, 2012.
about-fshungary; accessed 20.05.2013.
http://fshungary.hu/
[3] Rolf Fischer. Elektrische Maschinen 15. edition. HANSER, München, 2011.
[4] Kwang Hee Nam. AC Motor Control and Electric Vehivle Applications. CRC
Press, New York, 2010.
[5] Markus Hermwille. Igbt driver calculation. Technical report, Semicron,
2007.
http://www.semikron.com/skcompub/en/AN-7004_IGBT_Driver_
Calculation_rev00.pdf.
[6] SAE International. 2013 Formula SAE Rules, 2012. http://students.sae.
org/competitions/formulaseries/rules/2013fsaerules.pdf.
[7] Microsemi. Field oriented control of permanent magnet synchronous motors.
Technical report, Microsemi Cooperation, 2012. http://www.actel.com/
documents/SF_FOC_PMSM_HALL_UG.pdf.
[8] J. Pyrhönen, T. Jokinen, and V. Hrabovcová. Design of Rotating Electrical
Maschines. Wiley, Chichester, UK, 2008.
[9] Muhammad H. Rashid. Power Electronics Handbook. Academic Press,
2001.
[10] Wikipedia. Direct torque control, June 2013. http://en.wikipedia.org/
wiki/Direct_torque_control.
52
A
A.1
Matlab Code
Matlab code for square wave generation
freq = 1;
% Base Frequency
fs=freq*1000;
% Sampling frequency
A = 1;
% Amplitude
t = 0:1/fs:32/freq;
y = A* square(2*pi*freq*t);
figure(1);
clf(1,'reset');
plot(t,y);
axis([0 2/freq −1.2*A 1.5*A]);
xlabel('Time [s]');
ylabel('Voltage[V]');
title('SquareWave');
set(gcf,'color','w'); %set background to white
%set(gca,'XTick',0:pi/4:2*pi)
%set(gca,'XTickLabel',{'0','pi/4','pi/2','3/4pi','pi','5/4pi','6/4pi...
','7/4pi','2pi'})
% Fast Fourier Transform of Square Wave
L = length(y); %Length of Signal
NFFT = 2ˆnextpow2(L); % Next power of 2 from length of signal => FFT...
length
FFT = fft(y,NFFT)/L;
%FFT is symmetric => through away second half
FFT=FFT(1:NFFT/2+1); %+1 ?
%Take the magnitude of FFT
mFFT=2*abs(FFT);
%create frequence vector
f = fs/2*linspace(0,1,NFFT/2+1);%(0:NFFT/2−1)*fs/NFFT;
% Plot single−sided amplitude spectrum.
figure(2);
clf(2,'reset');
plot(f,mFFT);
title('Amplitude Spectrum of SquareWave');
xlabel('Frequency [Hz]');
ylabel(' | Voltage[V ] | ');
set(gcf,'color','w');%set background to white
axis([0 10*freq 0 1.2*A]);
53
A.2
Matlab code for modified sine wave generation
clf(1,'reset');
clf(2,'reset');
freq = 1;
% Base Frequency
fs=freq*10000;
% Sampling frequency
A = 1;
% Amplitude
n= 20;
%n Periods
%D = [2.5 10 17.5 20]' * 1/freq;
% pulse delay times
D= 0.25*1/freq:1/freq:n/freq;
D2=0.75*1/freq:1/freq:n/freq;
t = 0 : 1/fs : n/freq;
% signal evaluation time
w = 0.25*1/freq;
% width of each pulse
y = A*(pulstran(t,D,'rectpuls',w)−pulstran(t,D2,'rectpuls',w));
figure(1);
plot(t,y);
axis([0 2/freq −1.2*A 1.2*A]);
xlabel('Time [s]');
ylabel('Voltage[V]');
title('Modified Sinus Wave');
set(gcf,'color','w'); %set background to white
% Fast Fourier Transform of Square Wave
L = length(y); %Length of Signal
NFFT = 2ˆnextpow2(L); % Next power of 2 from length of signal => FFT...
length
FFT = fft(y,NFFT)/L;
%FFT is symmetric => through away second half
FFT=FFT(1:NFFT/2+1); %+1 ?
%Take the magnitude of FFT
mFFT=2*abs(FFT);
%create frequence vector
f = fs/2*linspace(0,1,NFFT/2+1);%(0:NFFT/2−1)*fs/NFFT;
% Plot single−sided amplitude spectrum.
figure(2);
plot(f,mFFT);
title('Amplitude Spectrum of Modified Sine Wave');
xlabel('Frequency [Hz]');
ylabel(' | Voltage[V ] | ');
set(gcf,'color','w');%set background to white
axis([0 10*freq 0 A]);
54
A.3
Matlab code for pure sine wave generation
sinfreq = 1;% Base Frequency
fs = sinfreq*10000; %sampling frequency
fc = 20;
%carrying frequency
A=1;
%Amplitude
t = 0:1/fs:32/sinfreq;
desired pos signal = A*sin(2*pi*sinfreq*t);
carrying signal = A*sawtooth(2*pi*fc*t,0.5);
desired vel signal = zeros(size(desired pos signal));
desired vel signal = sign(diff(desired pos signal));
PWM = zeros(size(desired pos signal)); %fill PWM with zeros
figure(1)
clf(1,'reset');
plot(t,carrying signal,'g');
hold on
for k= 1:length(desired pos signal)
if carrying signal(k) > desired pos signal(k)
PWM(k)=−1;
elseif carrying signal(k) < desired pos signal(k)
PWM(k)=1;
end
end
plot(t,desired pos signal)
hold on
plot(t,PWM,'r')
title('Pure Sine Wave PWM switching');
xlabel('Time [s]');
ylabel(' | Voltage[V ] | ');
axis([0 1 −1.2 1.2]);
set(gcf,'color','w'); %set background to white
annotation('textbox',...
[0.001 0.817 0.095 0.072],...
'String',{'Vdc'},...
'LineStyle','none');
annotation('textbox',...
[0.001 0.49 0.095 0.072],...
'String',{'Vdc/2'},...
'LineStyle','none');
y = PWM;
% Fast Fourier Transform
55
L = length(y); %Length of Signal
NFFT = 2ˆnextpow2(L); % Next power of 2 from length of signal => FFT...
length
FFT = fft(y,NFFT)/L;
%FFT is symmetric => through away second half
FFT=FFT(1:NFFT/2+1); %+1 ?
%Take the magnitude of FFT
mFFT=2*abs(FFT);
%create frequence vector
f = fs/2*linspace(0,1,NFFT/2+1);%(0:NFFT/2−1)*fs/NFFT;
% Plot single−sided amplitude spectrum.
figure(2);
clf(2,'reset');
plot(f,mFFT);
title('Amplitude Spectrum of sine wave');
xlabel('Frequency [Hz]');
ylabel(' | Voltage[V ] | ');
set(gcf,'color','w');%set background to white
axis([0 100*sinfreq 0 A]);
56
A.4
Matlab code for space vector PWM
function [S1,S2,S3,S4,S5,S6,sector,t1]=fcn(vd e,vq e,ramptime,...
theta e,Ts,Vdc)
S1=3;S2=3;S3=3;S4=3;S5=3;S6=3; %switching fault state values
%transform from synchronous into stationary frame
vd s=cos(theta e)* vd e−sin(theta e)* vd e;
vq s=sin(theta e)* vq e+cos(theta e)* vq e;
%find sector number in hexagon
if(vq s > 0)
if(abs(vq s) > (sqrt(3)*abs(vd s)) )
sector=2;
elseif(vd s > 0)
sector=1;
else
sector=3;
end;
elseif(abs(vq s) > (sqrt(3)*abs(vd s)))
sector=5;
elseif(vd s > 0)
sector=6;
else
sector=4;
end;
%calculate T1,T2 duties for each sector to generate relation between...
d,q
T1=sqrt(3)*Ts/(Vdc)*(sin(pi/3*sector)* vd s−cos(pi/3*sector)* vq s);
T2=sqrt(3)*Ts/(Vdc)*(−sin(pi/3*(sector−1))* vd s+cos(pi/3*(sector−1)...
)* vq s);
T0 =Ts−(T1+T2);
t1=[T0/4 T1 T2 T0/4 T0/4 T2 T1 T0/4];t1=cumsum(t1);
%sector I
if(sector==1)
pwmS1=[0 1 1 1 1 1 1 0];pwmS3=[0 0 1 1 1 1 0 0];pwmS5=[0 0 0 1 1...
0 0 0];
pwmS4=[1 0 0 0 0 0 0 1];pwmS6=[1 1 0 0 0 0 1 1];pwmS2=[1 1 1 0 0...
1 1 1];
for j=1:8
if(ramptime<t1(j))
57
break;
end;
end;
S1=pwmS1(j);S3=pwmS3(j);S5=pwmS5(j);
S4=pwmS4(j);S6=pwmS6(j);S2=pwmS2(j);
end;
% sector II
if(sector==2)
pwmS1=[0 0 1 1 1 1 0 0];pwmS3=[0 1 1 1 1 1 1 0];pwmS5=[0 0 0 1 1...
0 0 0];
pwmS4=[1 1 0 0 0 0 1 1];pwmS6=[1 0 0 0 0 0 0 1];pwmS2=[1 1 1 0 0...
1 1 1];
for j=1:8
if(ramptime<t1(j))
break;
end;
end;
S1=pwmS1(j);S3=pwmS3(j);S5=pwmS5(j);
S4=pwmS4(j);S6=pwmS6(j);S2=pwmS2(j);
end;
%sector III
if(sector==3)
pwmS1=[0 0 0 1 1 0 0 0];pwmS3=[0 1 1 1 1 1 1 0];pwmS5=[0 0 1 1 1...
1 0 0];
pwmS4=[1 1 1 0 0 1 1 1];pwmS6=[1 0 0 0 0 0 0 1];pwmS2=[1 1 0 0 0...
0 1 1];
for j=1:8
if(ramptime<t1(j))
break;
end;
end;
S1=pwmS1(j);S3=pwmS3(j);S5=pwmS5(j);
S4=pwmS4(j);S6=pwmS6(j);S2=pwmS2(j);
end;
%sector IV
if(sector==4)
pwmS1=[0 0 0 1 1 0 0 0];pwmS3=[0 0 1 1 1 1 0 0];pwmS5=[0 1 1 1 1...
1 1 0];
pwmS4=[1 1 1 0 0 1 1 1];pwmS6=[1 1 0 0 0 0 1 1];pwmS2=[1 0 0 0 0...
0 0 1];
58
for j=1:8
if(ramptime<t1(j))
break;
end;
end;
S1=pwmS1(j);S3=pwmS3(j);S5=pwmS5(j);
S4=pwmS4(j);S6=pwmS6(j);S2=pwmS2(j);
end;
% sector V
if(sector==5)
pwmS1=[0 0 1 1 1 1 0 0];pwmS3=[0 0 0 1 1 0 0 0];pwmS5=[0 1 1 1 1...
1 1 0];
pwmS4=[1 1 0 0 0 0 1 1];pwmS6=[1 1 1 0 0 1 1 1];pwmS2=[1 0 0 0 0...
0 0 1];
for j=1:8
if(ramptime<t1(j))
break;
end;
end;
S1=pwmS1(j);S3=pwmS3(j);S5=pwmS5(j);
S4=pwmS4(j);S6=pwmS6(j);S2=pwmS2(j);
end;
%Sector VI
if(sector==6)
pwmS1=[0 1 1 1 1 1 1 0];pwmS3=[0 0 0 1 1 0 0 0];pwmS5=[0 0 1 1 1...
1 0 0];
pwmS4=[1 0 0 0 0 0 0 1];pwmS6=[1 1 1 0 0 1 1 1];pwmS2=[1 1 0 0 0...
0 1 1];
for j=1:8
if(ramptime<t1(j))
break;
end;
end;
S1=pwmS1(j);S3=pwmS3(j);S5=pwmS5(j);
S4=pwmS4(j);S6=pwmS6(j);S2=pwmS2(j);
end;