Practical engineering

notes - part 1
Ciprian Antaloae

Topics include: Electric vehicles. High voltage electric drives. Electric motors
design & control. Power electronics and other related topics.
About

This is a collection of practical engineering notes hand-written over the period

2007-2011 during my postgraduate study in automotive electrical engineering.
It includes topics that I considered important to keep in mind over time.

This is Part 1, covering the rst 100 notes out of a total of about 500.

The purpose is to help students of electrical engineering and engineers working

on electric vehicles, by explaining concepts ranging from basic to advanced in a
way that is hopefully easier to understand than typical books or lectures.

Each hand-written note has a “So what?” section attached, where I aimed to
describe in what way the information is useful in a practical sense.

Bear in mind that for some of the topics, more advanced technology may have
been developed since the time of writing. I encourage you to take these notes as
a starting point for your own critical exploration of the eld.

Please go ahead and share with anyone that may nd this information useful.

Ciprian Antaloae

7 July 2023

Page 2 of 2
fi
fi
fi
1. Electric motor phase current signals sampling

• Phase currents are sampled by two ADCs (Analog to Digital converters) for
redundancy. A plausibility check of the sampled signals is performed before further
processing.

• TC179x represents the In neon TriCore 179 family of automotive processors.

• To detect stuck result-registers or multiplexed errors, a periodic rectangular signal is

fed to the ADC inputs with a time period of T=20ms.

So what?

• Electric motor control for Electric Vehicles is subject to very strict automotive safety
standards (e.g. ASIL C/D). To meet these standards, extra plausibility checks of the
signals coming from the 3 phase current sensors are needed. ASIL = Automotive
Software Integrity Levels ranging from A (least stringent, e.g. rear lights) to B, C and D
which is the most stringent (e.g. airbags).

Page 1 of 100
fi
2. Monitoring of electric motor phase currents

• As the sum of the phase currents is zero in theory, it is checked that this sum is not
bigger than a pre-set maximum constant. Otherwise, an error is generated.

• Redundant ADC register values are compared for the 3 phase currents.

• Each phase current is checked to be inside the range, as given by the calibration.

• Phase current signal offsets are also checked to be within the calibration range.

• A static and dynamic offset is added to the sensor gain.

So what?

• It is important to understand that the motor control software works with imperfect sensor
information. With respect to the 3 phase currents, the control software constantly runs a
number of checks on the values received from the sensors before using these in the
control algorithm itself. The ranges used for these checks are calibratable.

Page 2 of 100
3. Active short circuit (safe state)

• Either the low-side or the high-side IGBTs are turned on, to ensure no current is owing

So what?

• The drawing is a basic representation of a the standard DC/AC inverter (controller) for
electric motors, which includes 6 power switches (IGBTs = Insulated-Gate Bipolar
Transistors).

• In this diagram the 3 upper IGBTs are in an open state, the 3 bottom IGBTs are closed.

• Closing either the 3 upper or the 3 lower IGBTs creates the so-called safe state in an
electric drive, by interrupting the power ow between the battery and the electric motor.

• Electrically, what happens is that the 3 phases of the electric motor are connected
together (short-circuited, hence the name). This is typically not a problem for the electric
motor, but it is dependent on its magnetic design and also can be applied up to a
maximum speed.

• Active short circuit can be applied by the inverter itself as a defensive mechanism in case
of critical operating conditions (e.g. over-temperature or over-current), or the vehicle
control unit can send a request to the inverter to go into a safe state (e.g. in case of a
vehicle crash or abnormal/unknown operation).

Page 3 of 100
fl
fl
4. HVIL loop

• HVIL = High Voltage Interlock Loop

So what?

• This a simpli ed schematic of a HVIL which is implemented in vehicle high-voltage

systems for safety purposes.

• The HVIL is a low-voltage, low-current circuit which is integrated in the high-voltage

components and cables in the vehicle. When all these components are connected
together, the HVIL becomes an electrically closed circuit. When any one component is
disconnected (e.g. an orange power cable) the HVIL becomes an electrically open
circuit.

• This is how the vehicle controller checks if all the high voltage components are
connected, before allowing the vehicle to turn on.

• The speci c checking mechanism is based on a small current supply in the battery
(10-20 milliamperes) and of a voltage measurement in other components (e.g. inverter)
which is in the range of 4-6 Volts.

Page 4 of 100
fi
fi
5. Loop inductance

• It is the area loop size which directly affects the inductance, and less so the wire
diameter

• (Symbols) Circuit ground. Chassis ground. Digital ground. Analog ground.

• DC current path (least Resistance). High frequency current path (least inductance).

So what?

• Direct current (DC) ows in a closed circuit by taking the path of least resistance. For
alternating current (AC) current, the equivalent path is the one of least inductance even
if it may be longer for example.

• This is relevant to understanding printed circuit board (PCB) designs in general and how
high frequency AC currents may create unexpected or unintended operation. For wires,
AC current ows on the surface only - the more so as the frequency is higher (skin effect)

Page 5 of 100
fl
fl
6. Permanent magnet synchronous motor (PMSM) demagnetisation

• (A) reversible demagnetisation is proportional to the increase in temperature of the

rare-earth magnet material (e.g. Neodymium and Samarium lose 0.11% and 0.05% of
the magnetic eld strength per extra degC, respectively).

• (B) irreversible magnetisation can be caused by high opposing magnetic elds,

vibration or ageing. In motor control theory, as part of normal operation the motor
current component aligned with the magnetic d-axis can be applied with a negative
polarity to temporarily reduce magnetisation. If this negative magnetic eld is
accidentally too high (e.g.in case the rotor position sensor is not calibrated correctly), it
can lead to permanent demagnetisation of the magnets, making the rotor unusable.

So what?

• In electric drives, the motor torque is not measured but estimated. The variation of
magnet eld strength is directly relevant to this estimation of motor torque, so it is
important to know how the eld strength changes at high rotor temperatures.

Page 6 of 100
fi
fi
fi
fi
fi
7. Field-weakening demagnetisation

• The plots show the typical torque vs speed curves of PM synchronous motors

• This type of demagnetisation is reversible and is standard in motor control

• Beyond base speed (⍵b) the magnetic eld of the rotor has to be “weakened” in order
to allow the motor to continue to produce torque. The eld is weakened via a negative
current aligned with the magnetic axis (id). As a result of using “more id” to maintain a
constant magnet eld (and therefore a constant electromotive force e.m.f.), the other
current component producing torque (iq) has to be reduced as their vector sum is
limited by the battery supply current. Hence the torque decreases beyond base speed.

So what?

• It is standard motor control to apply negative id current beyond the base speed.

• It is interesting to understand that base speed is the point where the e.m.f equals the
battery supply voltage. Hypothetically, if the magnetic eld were not weakened beyond
base speed, the e.m.f. would continue to rise linearly with speed. Hence, the opposing
voltage generated by the motor (the e.m.f.) would become higher than the supply
(battery) voltage and no current could ow to the motor to produce any further torque.

Page 7 of 100
fi
fi
fl
fi
fi
8. Critical id which cancels the (magnetic) ux

• Here, critical id refers to the id value that creates an equivalent opposing magnetic eld
strength equal to the magnetic eld strength of the rotor permanent magnet (⍦PM)

• The magnetic ux produced by the stator current component id (⍦R) is the product of id
and the d-axis inductance Ld. Hence one can easily calculate the critical id value.

So what?

• The critical id is considered more of a limit rather than something that is used for active
control of PM synchronous motors.

• Essentially one does not aim to fully cancel the rotor magnetic ux. Instead, the target is
to apply the minimum id current needed in order to lower the e.m.f. and be able to
produce the requested amount of torque above base speed.

Page 8 of 100
fl
fi
fl
fl
fi
9. Critical id which cancels the (magnetic) ux

• PM synchronous motor dq model

• dq: refers to the two orthogonal axes, which form a reference system that is easy to
understand and use for alternating current motors.

• The d-axis is the magnetic axis of the motor, the q-axis the quadrature (orthogonal) axis
that is the only or main torque-producing axis (depending on the type of PM motor)

• The set of equations (1)-(4) characterise the base operation: (1)-(2) are the voltage
equations, (3) are the ux equations and (4) is the torque equation.

• Rs=motor stator resistance, Ld/Lq =d/q axes inductances, ⍵e =motor electrical speed,
p=number of the rotor magnetic pole pairs.

• To test for demagnetisation: set Id=Id=0 in the motor control system. For the control
system to follow these references, equation (1) has to hold i.e. Vq=⍦PM x ⍵ where all the
quantities are either known or measurable. If Vq>⍦PM x ⍵e then ⍦PM is lower than
expected which indicates demagnetisation of the rotor and a need for motor servicing.

So what?

• This set of equations is considered the base framework for understanding the
relationship between voltages, currents, ux and torque. It is used extensively in
everyday communication between engineers.

Page 9 of 100
fl
fl
fl
10. Star-delta comparison

• Electric motor stators can be con gured either in a star or delta connection

• Each con guration has its advantages and disadvantages - hence the design engineers
pick the suitable con guration for their application

• In star-connections, the phase voltage (Vph) is a sinusoid with the peak equal to the line
voltage (VL) divided by 1.73 (square root of 3). However, the phase current is equal to
the line current.

• The opposite is true for the delta-connections, where the phase voltage is equal to the
line voltage and the current is lower by a factor of 1.73.

So what?

• Star-connected motors produce higher peak torque due to the higher phase current,
but have a lower base speed and also a lower maximum speed due to the higher losses
at high speed.

• Delta-connected motors produce lower peak torque due to the lower phase current, but
have a higher base speed and a higher maximum speed due to lower losses at high
speed.

• One can think of the these two con gurations as the equivalent of the different gears of
a conventional transmission - which convert the same amount of available power into
higher or lower torque vs. speed. (See Note #12)

Page 10 of 100
fi
fi
fi
fi
11. Modulation index

• The modulation index (m) is a measure of how much of the battery voltage VDC is
actually applied to the motor windings (Vref), to maximise the power transfer.

• m ranges from 0 to 4/π. (4/π is approx. equal to 1.27 and is the maximum physically
possible with standard 2-level inverters).

• The most common modulation method for electric drives is SVM (space vector
modulation).

• The modulation index range is commonly re-scaled to 0-1. On this re-scaled range, the
SVM index is equal to 0.866, i.e. 86.6% of the battery voltage is being applied.

So what?

• The higher the modulation index, the more power is transferred from the battery to the
motor. But beyond a point (m=0.902), increasing m further is very “costly” in terms of
higher losses and electrical noise (higher order harmonics) and higher torque ripple.

• For applications where a smooth torque is needed, it is common to limit m to 0.902 and
give up the extra power.

Page 11 of 100
12. Star-delta torque-speed envelope

• See Note #10 for a description of star-delta motor winding connections

• This plot shows the torque-speed envelope of an electric motor, where everything is
kept the same apart from the type of connection (start or delta)

So what?

• This is just a simplistic view of the torque-speed differences between start-delta. In

reality, the differences include a deep analysis of the electromagnetic, thermal and
mechanical implications

• For the higher torque of the star connection, the motor design is more challenging due
to higher electromagnetic elds (to avoid magnetic saturation), higher losses at peak
torque and possible packaging constrains (creating a connection of the phases together
to a neutral point)

• Start-delta should be considered as one of the rst motor design criteria, based on the
main requirements of the target application and the speci cation of the battery and
inverter.

Page 12 of 100
fi
fi
fi
13. Inverter frequency derating

• Derating of power in electric drives is common and can be caused by a number of

factors, including the frequency of the AC currents (see also Note #19).

• When the motor speed is very low (e.g. <500rpm), the motor current frequency is
correspondingly low (in synchronism). For the inverter to create the lower frequency AC
currents, the 6 power switches have to be controlled to be in the ON/OFF state for
longer continuous time periods.

So what?

• Longer ON/OFF periods of the power switches creates higher temperature variations,
which is the main contributing factor to the lifetime of the power switches.

• In order to preserve the lifetime of the power switches, the inverter software limits the
amplitude of the current transferred from the battery to the motor (derating). A lower
current amplitude limits the temperature swings during the longer ON/OFF states and
hence limits the impact to the lifetime of the inverter.

Page 13 of 100
14. Middlebrook criterion

• The Middlebrook criterion states that if the output impedance of the input lter is above
the input impedance of the load or connecting component (in the electric circuit), this
may lead to current and/or voltage oscillations.

• As a design rule of thumb, to avoid possible oscillations in a circuit including 2

connected components: the output impedance of the rst component should be below
the input impedance of the component it connects to (across the operating frequency
spectrum).

So what?

• This theoretical criterion should be used a rule of thumb in the design of electrical and
electronic circuits. Also useful when troubleshooting oscillations of voltage or current,
observed in between connecting electrical components.

• For electric drives, where more and more auxiliaries are connected directly to the high
voltage 400/800V bus (e.g. A/C compressors, heaters etc), oscillations can occur at the
connection points which can have an effect on the whole electric system

Page 14 of 100
fi
fi
15. The modulation index dependency of 3-phase connection

• The modulation index (m) is independent of the type of motor connection (see also
Note #11 for more information on the modulation index).

• An easy to understand de nition of m = ratio between the peak of the fundamental

harmonic voltage (being applied to the electric motor windings) and the peak of the
fundamental harmonic of six-step commutation.

• Six-step commutation applies 6 “voltage blocks” per electrical period to the motor
windings. The fundamental frequency component of this waveform has a peak value of
2/π.
• Important to understand that these concepts are related to the main (fundamental)
frequency component of the phase voltage. The phase voltage applied to the motor
windings, also the six-step commutation, include other higher order harmonics as well.

So what?

• The modulation index and inverter modulation strategy are equally applicable to star/
delta-connected motors, with similar effects to the output power.

Page 15 of 100
fi
16. Modulation index (of inverters, for motor control)

• The index of the 3 main types of inverter modulation: Space Vector Modulation (SVM)
0.866, Sinusoidal PWM 0.76, SVM with over-modulation 1.

• The “block” 6-step voltage pro le (line-line inverter voltage) corresponds to full
modulation and follows the hexagon pro le in the SVM vector representation (index of
4/π - see Note 11). The circle inscribed in this hexagon is a representation of the
sinusoidal waveform of this voltage (index of 0.76 as above).

So what?

• Its is useful to understand the different possibilities for index values and to visualise the
corresponding line-line voltage waveforms created - leading to a compromise between
power delivered to the motor and the losses and NVH (noise, vibration, harshness)
effects on the motor.

Page 16 of 100
fi
fi
17. Overmodulation (of inverters, for motor control)

• Looking at the typical motor torque vs speed pro le (as in plot above), the
overmodulation extends the base speed (shifts the corner point to the right). It improves
the torque in the eld weakening region.

So what?

• The topic of overmodulation (also described in Notes 11 & 16) refers to increasing the
peak power of the motor, and not the peak torque.

• The peak power is the corner point on the torque-speed pro le, where the speed is also
called base speed.

• Overmodulation extends the base speed to a higher level. In the same way, a decision
can be made to lower the modulation index, in which case the base speed level moves
to the left.

• Depending on the application, the engineering team can decide to either increase the
modulation in order to gain more peak power at the cost of higher losses and NVH, or
the decision can be made that NVH should be improved at the cost of reducing the
peak power.

• Applications where high power/ NVH is preferred include vehicle traction motors (up to
an extend) or industrial motors.

• Applications where low NVH /power is preferred include electric power steering or
active suspension, where the motor operation should be very smooth.

Page 17 of 100
fi
fi
fi
 18. IGBT temperature measurement

• IGBTs temperature measurement - the individual IGBT die temperature is not typically
measured (although a technical solution can be to measure it using the diode
connected in parallel to the IGBT die)

• Instead, the temperature of the DCB plate is measured as an average of multiple IGBT
dies temperature.

• DCB=direct copper bonded, the plate or substrate in the overall power module where
the IGBTs are mounted and connected - typically made of ceramic material.

So what?

• Ideally one would want to measure directly the temperature of each IGBT die, to have
exact information about its state of health and possible failures that can occur.

• However, measuring the individual die temperatures is impractical because of increased

cost and increased complexity of the design and restricted packaging volume.

• Also one should bear in mind that in practice, “one” IGBT switch (as seen in the 6-switch
simpli ed inverter con guration) is made up of multiple IGBT dies connected in parallel.
Hence the topic of measuring individual die temperature becomes even less attractive
and practical, the more dies are jointly connected together.

Page 18 of 100
fi
fi
19. Current derating (of inverters, for motor control)

• Inverters limit the peak current to the motor (derating) based on either one or more of
the following factors:

1. DCB temperature level (see Note 18 for DCB)

2. DC link (HV battery) voltage level (because the power dissipation of semiconductors
is dependent on the switched voltage). The higher the voltage, the higher the losses
and the more this can lead to a reduction in the allowed inverter output current.
3. Frequency of AC currents derating (because the IGBTs heat up faster at low motor
current frequency - the reason being that under this condition the IGBTs are on for
longer at a time).
4. Modulation factor (see Notes 16, 11). The higher this is, the higher the losses and
the higher the resulting temperature rise on the inverter, leading to current derating.
5. Time derating - (a) low current frequency, (5) high current frequency (linked to #3)
6. Cooling pump or cooling circuit in general failure.
So what?

• The common denominator across all points is a maximum temperature limit that the
IGBTs and the power module of the inverter should not exceed. In order to achieve this,
the output current is proportionally reduced. This has an effect on the vehicle drivability,
practically can lead to inconsistent or low power output for the same driver pedal
request. Early EVs had the issue of inconsistent acceleration times after the rst 10 times.

Page 19 of 100
fi
20. Inverter speci c power

• DC/AC inverter speci c power is equal to machine power x 1.6 to account for the power
factor and ef ciency losses.

• The speci c power is measured in VA (Volt x Amp), also known as apparent power.

So what?

• This is just a rough indication or rst order rule of thumb approximation of the VA rating
of an inverter, compared to the motor it is driving.

• The actual ratio can be different for different motors, for different ways of controlling the
same motor etc.

• The key implication is that the higher the VA rating compared to the motor power (W),
this is an indication that the motor requires a large amount of reactive power (VAR= VA
reactive) in order to run.

• A high VAR means large reactive power circuit components are needed on the DC link
(the electrical connection between the HV battery and the inverter), resulting in high
packaging volume and high material cost.

• The main VAR component is the DC link capacitor, which (depending on the build and
material technology) can be voluminous and costly.

• Important to note that a 0VAR can be targeted and controlled by the software of the
inverter, but in this case the motor peak power would be reduced. The method is called
Unity Power Factor (UPF) and controls the motor phase voltages and currents to be
always in-phase.

Page 20 of 100
fi
fi
fi
fi
fi
21. Motor CPSR (Constant power to speed ratio)

• A low CPSR (Constant Power to Speed Ratio) leads to an inverter design for high
current, hence a high size (volume)

• A high CPSR (e.g. 6:1) is desirable (from an inverter design point of view).

So what?

• The plot shows a typical torque vs speed motor curve, where the area to the right of the
base speed (see Notes 7, 17) is the motor constant power area. A high CPSR ratio means
that the maximum speed of the motor is high compared to its base speed.

• From an inverter design point of view, to keep the material cost and the packaging
volume low, it is best to aim for a low current design.

• A low current design of the inverter means that the area of constant peak torque of the
motor is quite short (low base speed), hence the rest of the curve (above base speed)
should be a lot larger (i.e. high CPSR)

• This idea is mostly theoretical, since in practice the requirements of the application
determine the torque-speed curve of the motor, and these requirements are then
cascaded to the inverter in terms of voltage, current ratings etc (and not the other way
around).

Page 21 of 100
 22. Sinusoidal vs. Brushless motors

• Sinusoidal PMSM: parallel magnetisation

• Brushless DC: radial magnetisation, higher RMS current —> higher power density

• Brushless DC: current is injected into one phase and extracted in the next phase—> one
of the 2 phases can be used as a position sensor (not possible in PMSM)

So what?

• The sinusoidal PMSM (Permanent Magnet Synchronous Motor) has a sinusoidal back-
emf waveform (the induced voltage in the motor windings from the rotating magnetic
eld generated by the permanent magnets of the rotor).

• The brushless DC motor (which is also actually an AC motor) has a trapezoidal back-emf
waveform. Due to the working principle, this type of motor has a higher power density
than the sinusoidal PMSM and is the preferred option where high power is needed and
a moderate-high level of NVH can be tolerated (e.g. industrial applications).

• Due to the operating principle of the brushless DC motor, where current is present only
in any 2 out of the 3 phases at any time, this aspect can be used to determine the
position of the rotor. The advantage is that a position sensor is no longer needed, thus
saving material cost.

• Note: the name brushless DC comes from the traditional brushed DC motor, where
mechanical brushed were used to switch the direction of the current.

Page 22 of 100
fi
23. Cogging torque

• Cogging torque - caused by an un-even air gap permeance resulting in the magnets
constantly seeking a position of minimum reluctance.

• Cogging torque = no current (0A) torque

So what?

• There are two types of motor torque - excitation torque, created by a given controlled
current, and non-excitation torque or cogging torque, which is the natural physical effect
of how the permanent magnets are installed and positioned inside the motor rotor.

• Cogging torque always sits on top of the excitation torque, creating a certain torque
ripple effect.

• The aim of motor design is to keep the cogging torque to a minimum, in order to have a
minimum torque ripple.

Page 23 of 100
 24. Maximum power and voltage transfer (in electric circuits)

• For maximum power transfer, the impedance of the source should be equal to the
impedance of the load

• For maximum voltage transfer, the the impedance of the source should be low
compared to the impedance of the load

So what?

• This idea refers to any electric circuit - from industrial applications, to traction electric
drives in vehicles where the source is the HV battery and the load is the inverter + motor
combination.

• The typical applications where this concept is relevant include power converters or
lters. It can be also of use in electric drives, when comparing the source and load
impedances (2 Bode plots, as a function of frequency), in order to understand how
power and voltage are being transferred.

• Final note: in electric drives, during regenerative braking, the motor + inverter pair
become the source and the HV battery the load. In order to understand how much
power can ow back into the battery, one should also take into consideration again the
relationship of the 2 impedances, at the frequency range of operation.

Page 24 of 100
fi
fl
25. Motor synchronous speed

• Motor synchronous speed (rpm)= 120/n x f (Hz)

So what?

• As the name indicates, this formula applies only to synchronous electric motors

• n = number of magnetic poles of the motor rotor.

• In practice, this equation is useful to work back from the motor speed in rpm, to the
corresponding frequency of the motor currents for this particular speed.

• Once the motor frequency is known, one can investigate or troubleshoot the operation
of the electric drive for that particular point or its immediate frequency range.

• For example, if the electric drive exhibits a particular resonance at a speci c speed point,
this is translated in the frequency domain rst before starting the investigation.

Page 25 of 100
fi
fi
26. MOSFET - main types

• MOSFET = Metal Oxide Semiconductor Field Effect Transistor

• MOSFET types - enhancement type and charge-depletion type

• Ve = voltage applied across the gate-source to control the switch on/off of the MOSFET

• P-Channel and N-Channel are turned on/off differently, as per the table.

So what?

• Useful to understand the different types of MOSFETS, even though typically the N-
channel enhancement type is the most widely used variant in automotive applications.

• Useful when inspecting circuit diagrams, to understand the logic of which MOSFETs turn
on /off under the given Ve voltage created by the gate circuitry.

• The charge-depletion type of MOSFET is also considered to be permanently ON, unless

turned off by the Ve voltage applied to its gate.

Page 26 of 100
27-28. Reluctance torque

• Reluctance torque formula (highlighted area of the equation)

• Reluctance is to a magnetic circuit the same as the resistance to an electric circuit.

So what?

• In Note 23, the term excitation torque was described. Reluctance torque is one
component of the excitation torque

• For synchronous motors (especially PMSMs), the main component of the excitation
torque is produced by the permanent magnet eld interacting with the q-axis current of
the stator, producing a coupling magnetic eld which generates the torque in the motor
air gap (see Note 9)

• Other types of motors (synchronous reluctance or switched reluctance for example),

produce only this type of reluctance torque.

• Even though the term reluctance conceptually indicates a resistance effect, in practice
one can think of reluctance torque as the effect created by the natural tendency of the
magnetic ux circuit in the motor to reach its state of equilibrium - which is to minimise
the length of the magnetic eld circuit. In this transition to nd the natural equilibrium,
certain pulling forces are created between the stator and rotor which add to the total
torque created in the air gap of the motor.

• In the same way as currents ow in loops, magnetic ux also requires a closed loop. All
the media that the ux lines cross (magnetic material, air etc.) opposes a speci c
resistance to the ow of the ux - this resistance is called reluctance and is the analog of
the resistance that opposes the current ow in an electric circuit.

Page 27 of 100
fl
fl
fl
fi
fl
fl
fl
fi
fi
fl
fi
fi
- blank page -

Page 28 of 100
29-30. Magnetic ux and reluctance

• Flux = MMF/ Reluctance (Magnetic Ohm’s law)

• B=magnetic ux density, unit =T (Tesla). MMF= Magnetomotive force

• The plot shows the nonlinear relationship between reluctance and magnetic ux
density, and the so-called Saturation area above approx. 1.8T

So what?

• The target when designing any magnetic circuit is to maximise the Flux Density for the
given circuit physical characteristics (length, material etc) and the input magnetising
current

• In order to maximise the Flux Density, the reluctance should be as low as possible (as
per the equation above). However, the reluctance is in turn dependent on the Flux
Density itself.

• In magnetic soft materials (easily magnetisable materials), one can apply a certain
amount of Flux Density up to a point, beyond which the reluctance increases sharply -
making the operation in this area very inef cient and therefore to be avoided. This is also
called the magnetic saturation area. The main takeaway point is that there is a maximum
Flux Density level to be observed, this depends on the application and materials.

Page 29 of 100
fl
fl
fi
fl
- blank page -

Page 30 of 100
31. Electromotive force (EMF)

• Electromotive Force (EMF) is caused by motion, by the “cutting” of the ux lines by the
moving conductor.

• EMF= - derivative of Flux vs time, where Flux = L x I (Inductance x Current)

So what?

• Basically EMF is an induced voltage, created by a motion action

• In electric drives, this concept applies to the electric motor and the fan that the
movement of the rotor magnetic eld over the motor stator windings induces an EMF
into the windings.

• If the rotor is not magnetised, then there is no induced voltage in the stator windings.
The rotor can be either permanently magnetised or via separate current excitation, it
does not matter.

• The EMF in electric motors is a sinusoidal per-phase voltage, which has a variable
synchronous frequency (see Note 25) and a variable amplitude proportional with the
speed.

• The EMF in electric motors is out of phase (opposes) the applied phase voltage, hence
during normal operation it causes the amount of current that can ow into the windings
to be lower.

• The base speed point of a torque-speed motor curve is where the EMF amplitude is
large enough that it becomes equal to the battery voltage minus the current x resistance
of the windings.

• Beyond the base speed point, the EMF amplitude continues to rise, but the control
software starts to keep this amplitude constant by the application of de-magnetising
current. This is why the torque also starts to decrease, because less & less current is
available for torque creation as the speed increases.

Page 31 of 100
fi
fl
fl
 32-33. iPMSM d-q inductances

• Normally in an iPMSM (interior PMSM) Lq>Ld because the effective air gap of the d-axis
is much larger than the one on the q axis. (see Note 9 for description of d,q-axes)

• The d-axis reluctance is greater than the q-axis reluctance because the permeability of
high ux density permanent magnets is that of air —> essentially the magnet thickness
extends the air gap (between stator and rotor).

So what?

• Interior PMSM - PMSM where the rotor magnets are embedded inside the rotor (as
opposed to being mounted on the surface of the rotor)

• Looking at a typical arrangement of magnets in a PMSM (sketch above), the d-axis is

aligned with the centre of the magnet and the q-axis is 90 electrical degrees away. From
a magnetic point of view, the magnets act as an air gap - this is the reason the reluctance
on this d-axis is greater than on the q-axis, and hence the d-axis inductance is lower.

Page 32 of 100
fl
• In Note 27 the topic of reluctance torque was described. In the equation, since Ld<Lq
then the term (Ld-Lq) is negative. Also, typically the Id current is negative, hence the
resulting reluctance torque value is positive and contributes to the overall torque
produced by the motor.

• In practice, there is a well de ned strategy for how to create the optimum amount of
reluctance torque, which starts with how the magnetic circuit of the motor is de ned -
with the aim to have a low Ld and a high Lq, to boost the reluctance torque creation and
minimise the amount of magnet material to be used for the excitation torque (to lower
the material cost).

Page 33 of 100
fi
fi
34-35. Symmetrical square waves

• Symmetrical square wave formula. Here I refers to a current waveform, but this is
generally applicable to any symmetrical square wave.

• In = the n-th harmonic of the wave (n=1 being the fundamental)

• T is the wave time period. d is the duty cycle from 0 to 1. Tr represents the rise & fall time
(due to symmetry). For d=50%, only odd harmonics are present - 1,3,5 etc

• Harmonic spectrum of a square-like wave shown. There is a knee point at the frequency
(1/(π * tr ) where the amplitude of the higher harmonics becomes attenuated at a rate of
40dB per decade (logarithmic scale)

So what?

• In electric motor drives, the voltage waveform applied over the motor windings can be
symmetrical or not - depending on the modulation method (e.g. symmetrical PWM).

Page 34 of 100
• For symmetrical modulation algorithms, it is useful to understand the harmonic
spectrum of the voltage, to get an idea of any electrical, mechanical and
electromagnetic noise induced by the voltage harmonics into the motor itself.

• The frequency range for this topic is higher than the motor AC (synchronous frequency) -
from tens of kHz to MHz range. Hence, these could also play a role in electromagnetic
interference with other electronic equipment in range (e.g. vehicle radio or other
electronic devices and controllers).

Page 35 of 100
36. Luenberger observer

• Luenberger observer diagram

• Block representation of the main transfer functions (s - Laplace frequency domain), from
the Command Pc(s) to the Plant, Sensor (Physical system) and the corresponding
models for each in the Observed (modelled) system.

• There is an observer compensation block that links the Physical system and the
Observed system.

So what?

• The Luenberger observer adds a correction (compensation) to the modelled system,

such that the calculated (observed) output matches the reals sensor output.

• One purpose is to have a modelled system that is as representative as possible of the

real system - useful for systems that are not able to be fully de ned. Hence the name
observer - the system is being observed.

• The observer also provides as an output the observed or calculated sensor output - this
can be used for applications where the real sensor output is not satisfactory for the
application, e.g. too low bandwidth or not accurate enough. In these cases, both the real
and the calculated sensor outputs are used together in the application - via a dedicated
logic.

Page 36 of 100
fi
37. Computationally-ef cient Luenberger observer

• Computationally-ef cient Luenberger observer (in the Laplace frequency domain)

• The classical Luenberger observer (Note 36) is known to exacerbate sensor noise.

So what?

• This form of the Luenberger observer re-arranges and combines the operations of the
classical observer structure (Note 36) in order to reduce the computational effort at
runtime.

• Useful for real time applications where the number and complexity of computations
needs to be minimised.

Page 37 of 100
fi
fi
 38. Bene ts of DTC vs. Voltage-fed PWM such as SPWM, SVPWM

• Bene ts of DTC (Direct Torque Control, for motors) vs. Voltage-fed PWM such as SPWM
(Sinusoidal PWM), SVPWM (Space-Vector PWM):

A. Improved motor dynamic torque response

B. No inrush current

C. No sensitivity to DC voltage variation

• DTC is implemented using hysteresis band current control. The only type of motor
control which is current-based. The others are voltage-based (SPWM, SVPWM).

So what?

• Standard automotive motor control for electric drives is voltage-based (SPWM, SVPWM)

• DTC offers the fastest torque response, but voltage-based methods still meet the
requirements of automotive applications with response times of under 20micro-sec.

• HV battery voltage uctuations are no longer a concern for modern batteries, which hold
a relatively stable voltage across the state of charge range.

• However, it is useful to be aware in general of other methods to control electric motors.

Page 38 of 100
fi
fi
fl
39. The hysteresis band current control

• The hysteresis band current control is implemented with a Schmitt trigger, which are a
logic equivalent to latches.

• When the signal to be controlled (here denoted as u) is lower than the preset lower
threshold, the switch changes output to the LOW state (or OFF state)

• When the signal to be controlled is higher than the preset higher threshold, the switch
changes output to the HIGH state (or ON state).

So what?

• The Schmitt trigger was invented in 1934 by Otto Schmitt.

• The output maintains its value until the input changes below or above the given lower
and higher thresholds, respectively.

• DTC control (Note 38) is an application where Schmitt triggers can be applied, although
in the author’s experience these are not present in automotive electric drive
applications.

Page 39 of 100
40. Hysteresis band current control

• Hysteresis band current control

• It has a similar m (modulation index) to Sinusoidal PWM (0.7855) (in motor control)

• The smaller the hysteresis band, the higher the switching frequency of the inverter (Fs),
and the better the waveform (here, better = closely following the current reference)

• Unlike SPWM and SVPWM, hysteresis control leads to variable switching frequency (as
the input voltage changes for example)

So what?

• Hysteresis controllers are not common in modern electric drive applications. Instead,
constant-frequency modulation methods are preferred (e.g. SPWM, SVPWM etc).

• The general topic of hysteresis control could be useful for other applications, especially
where very speci c or atypical current waveforms are needed.

Page 40 of 100
fi
41. Hysteresis current control PWM

• Hysteresis current control PWM characteristics:

• Switching frequency changes with load

• The frequency of the predominant harmonic in the output voltage is approximately

equal to the switching frequency of the converter

• Hence, if the switching frequency Fs changes, the frequency of the predominant

harmonic changes

• Therefore, a lter to eliminate this harmonic can be dif cult (to design).

So what?

• Along the same lines of the previous Note, the challenge with the changing output
frequency is highlighted - here from the perspective of having to design a lter for this
frequency.

• This topic is also applicable to power converters for any power application.

Page 41 of 100
fi
fi
fi
 42. A good PWM technique should..

• A good PWM technique should:

• Turn on/off only one switch at a time

• Line-Line voltage (V_LL) should be switched not directly from + DC voltage to - DC

voltage, but in half steps from 0 to either +/- DC voltage

• Not contain or generate any subharmonics

So what?

• Normally it is dif cult to perfectly align the turn on/off of the PWM switching edges -
hence the rst comment of only turning on/off one at a time. For example, in classical 6-
switch inverters, in symmetrical PWM, only one inverter leg (vertical pair of 2 IGBT
switches) turns on/off at any given time.

• Applying voltages in steps from 0V to -/+ DC voltage is preferred to keep the electric
stress low on the load and to limit the electromagnetic noise generated by large voltage
swings (such as from -DC voltage to +DC voltage).

• Subharmonics (harmonics below the fundamental frequency of the application) or inter

harmonics (non-integer multiples of the fundamental) are dif cult to identify and hence
lter out, and invariably lead to unintended operation effects and extra losses.

Page 42 of 100
fi
fi
fi
fi
43. Multi-level inverters for high power applications

• Multi-level inverters are used in high power applications. In typical 2-level inverters, one
can only work with 2 levels of voltage: +DC voltage and -DC volatge. In multi-level
inverters, multiple voltage levels are available - to better synthesise the sinusoid.

So what?

• The drawings are conceptual representations of what multi-level voltage levels could be
- e.g. in steps of 1/2 DC voltage via the circuit schematic shown.

• In automotive electric drive applications, the complexity and cost of multi-level inverters
would not justify its adoption. The main application is industrial level power inverters of
1,000V+.

Page 43 of 100
44. Voltage source inverters (VSI)

• A voltage source inverter (VSI) is a 2-quadrant inverter: voltage always positive, current
can be positive or negative.

So what?

• The 4 quadrants refer to the voltage-current positive/negative combinations.

• In a VSI, voltage is applied to the load in a controlled way, to obtain the desired current
waveform for the application. One controls the current indirectly, via the pulse
modulation of voltage.

• An electric drive in vehicles is a VSI type application.

Page 44 of 100
45. Power transformers

• The voltage ratio in a transformer (E1/E2) is equal to the ratio of the number of turns on
the 2 sides of the transformer (T1/T2).

So what?

• Just a fundamental principle of operation for transformers in the linear magnetic

operating region. Saturation can occur (see Note 30).

Page 45 of 100
46. Six-step commutation

• For six-step commutation (6-switch inverter), the peak of the fundamental is 2/π * VDC

• For H-bridge commutation (4-switch inverter), the peak of the square wave fundamental
is 4/π * VDC. The diagram shows a basic representation of the H-bridge & transformer.

So what?

• The fundamental refers to the 1st sinusoidal harmonic of the resulting square-like wave
resulting from the switching of the DC voltage. This square-like wave can be six-step (see
Note 16) where there are 6 steps in a period of 360 electrical degrees, or can be two-
step for inverters with only 4 switches as in the diagram above.

• At the time of writing the note, it was not clear where the respective amplitudes of 2/π
and 4/π came from. In fact, these come from the Fourier frequency de-composition of
the square-like waves - where these numbers are the amplitudes of the 1st harmonics.

• It is important to understand the amplitude of the fundamental harmonic, as this is the

one converting into a meaningful physical output (e.g. torque for electric motors),
whereas the higher harmonics typically convert into losses and create electrical and
mechanical noise.

• Even though the six-step fundamental is 1/2 of the H-bridge fundamental, it is part of a
3-phase system where the total power is multiplied by 3 and additional voltage can be
transferred to the motor via speci c PWM methods (see Note 11): 15% - 27% more.

Page 46 of 100
fi
47. Induction and synchronous motors

• The induction motor always runs at a speed which is slightly lower than the synchronous
speed in motor mode - also called positive slip (angle).

• Synchronous speed = frequency of the stator ux.

• In generator mode, the rotor speed is slightly higher than the synchronous speed- also
called negative slip (angle).

So what?

• This is the reason why the induction motor is also called asynchronous - because the
rotor speed and the stator ux speed are not in-sync.

• The out-of-sync is deliberate and precisely controlled by the motor controller, its
purpose is to induce an additional current into the rotor, which creates the rotor
magnetic eld (same type of eld as generated by the permanent magnets in a PMSM).

Page 47 of 100
fi
fl
fi
fl
48. Induction motors time constant

• In an induction machine, the time constant of the stator ux is much lower than the time
constant of the rotor speed. Bringing the stator eld frequency down quickly leads to
the IM (Induction motor) operating as generator until the rotor speed decreases.

So what?

• In this context, the time constant refers to the time constant of the 1st order type of
electric circuit composed of the equivalent resistance and inductance values of the
stator and rotor. For the stator, this time constant is much lower and hence the voltage
applied creates a much faster response (rise time) of the current in the stator (compared
to the rotor).

• This difference in time constants can be used to obtain various control effects for the
induction motor, such as the sudden change in stator eld frequency leading to a
braking effect from the motor (generator). In this case, the rotor will also slow down and
aim to align its own speed to that of the stator eld (natural equilibrium), but in order to
keep the motor operating as generator, the stator eld frequency will continue to be
controlled to become lower and lower and not achieve synchronism.

• Synchronism of stator eld frequency and rotor speed in an induction motor means zero
torque - hence not an operating point to be actively controlled. For zero torque in
induction motors, the voltage is simply turned off (unlike PMSMs, where a small voltage
is needed to cancel out the torque created by the MMF induced by the rotor eld).

Page 48 of 100
fi
fi
fi
fi
fi
fl
fi
49. Induction motors line-line voltage

• In an IM (induction motor), the line-line voltage is a series of pulses and the line-line
current is a sinusoid (same as in other motors actually, where the current is sinusoidal)

• The fundamental frequency of the voltage is the frequency of the sinusoidal current. All
other harmonics on the voltage need to be high, to be ltered out by the IM inductance.

So what?

• Different PWM methods create different harmonic spectra (See Notes 34,35,42)

• If possible, one should choose the PWM methods that create the needed fundamental
and the "by-product” harmonics are much higher in frequency (as shown in the diagram
plot above, conceptually), such that these higher harmonics are easily ltered out by the
motor inductance.

• This is a nice theoretical concept, in practice once cannot create such a clean frequency
spectrum where all the unwanted harmonics are pushed to the higher end for ltering.

Page 49 of 100
fi
fi
fi
50. Voltage source inverters (VSI) and Current source inverters (CSI)

• VSI (voltage source inverter) - voltage does not reverse (always positive), because of
high DC link capacitance. Current does reverse (positive & negative) - see Note 44.

• CSI (current source inverter) - current does not reverse, because of high DC link
inductance (not current as in the text). Voltage does reverse.

So what?

• This is a conceptual note, to understand the fundamental operating differences of VSI

and CSI circuits.

• When thinking of the 4-quadrant operation of electric circuits (Voltage vs Current), the
VSI operates in the 2 quadrants of positive voltage and the CSI in the 2 quadrants of
positive current.

• In any electric circuit, capacitance “opposes” voltage and inductance “opposes” current
from a dynamic (rise & fall times) point of view.

Page 50 of 100
51. Current source inverters (CSI)

• CSI - power switches are different from VSI, the diode across the switch needs to be
removed to avoid the reverse voltage.

• CSI power switches need to block the reverse voltage: GTO, SCR (thyristors)

• CSI with IGBTs - schematic shown. Requires extra series diode (the one pointing down)

So what?

• Reverse voltage across a power switch is normal in a VSI with IGBTS. - such as electric
drives. It is part of the normal operation and is enabled by the anti-parallel diode across
the transistor.

• GTO = Gate Turn-Off Thyristor, SCR=Silicon Controlled Recti er (also a Thyristor) - these
are typical equivalents of IGBTs for CSIs.

• The same operation as a thyristor can be achieved with an IGBT and an extra series
diode blocking the reverse current.

Page 51 of 100
fi
52. Current source inverters (CSI)

• Selected comments: CSI - there must always be a closed path (from source to load),
because of the large inductor in the current source. For inductive loads (e.g. motors) a
slower turn on/off transition is required - hence lower switching frequency than VSIs.

So what?

• A CSI cannot be used for electric drives where under certain safety scenarios the source
and load have to be disconnected during operation. Also the lower switching frequency
makes them less desirable, as it creates higher motor NVH.

Page 52 of 100
53. Voltage source inverters (VSI)

• In a VSI, a high leakage inductance is preferable because the harmonic current is equal
to (see equation - where the leakage inductance appears at the denominator).

• As the leakage inductance is higher, the respective current harmonic IH is lower

(neglecting the resistance in the circuit). VH is the harmonic voltage and FH the
frequency of the harmonic.

So what?

• Leakage inductance refers to imperfectly coupled magnetic circuits creating losses in

the application. Hence this is not a desirable quantity as it “wastes” energy.

• However, there is a side-bene t of having high leakage inductance, that it reduces the
amplitude of the current harmonics which in turn improves the electrical ef ciency of the
application and lowers the NVH.

Page 53 of 100
fi
fi
54. Voltage and Current source inverters (VSI, CSI)

• CSI - short circuit proof. Rugged circuit. VSI - protection from short-circuit required.

• CSI - suitable for high power applications, but absolutely needs a closed-loop load
unlike the VSI.

• CSI - main line current harmonics: 5th, 7th, 9th, 13th. PWM not popular because the
power devices (thyristors) are slow to turn on/off.

So what?

• The VSI application for electric drives includes fuses on the DC link (between the battery
and the inverter) to protect both the battery and the load from a short circuit. This is one
disadvantage of a VSI compared to a CSI.

• The VSI has its own speci c current harmonics, some are linked to the application and
some generated by the PWM technique used. Since PWM can be deployed at high
frequencies (10-20kHz for standard IGBTs), advanced methods can be used where some
harmonics can be eliminated or compensated (depending on the requirement of the
application).

Page 54 of 100
fi
 55. The RL circuit

• The basic resistive-inductive circuit with an AC source.

• Z is the circuit impedance, as a function of the angular frequency ⍵ (rad/s)

• φ is the phase shift between the AC voltage and AC current in this circuit

• is the time constant, referring to how long it takes the current in the circuit to respond
to a voltage applied (or vice versa, if the source is a current source).

So what?

• This is a fundamental concept but it can come in handy when analysing the electric
motor operation and the basic building blocks of each phase - resistance, inductance
and a voltage source.

• For example, since the impedance is a function of frequency of the motor, and since the
source voltage is constant, these concepts together indicate that less and less current
can be controlled onto the motor at high speeds. There are other physical aspects
contributing to the same effect, such as the back-emf (see Note 31).

Page 55 of 100
𝜏
56. 3-phase system with & without neutral

• In a 3-ɸ, 4-wire connected motor, the fundamental sinusoidal currents add to zero and
the 3rd harmonics add up and ow via the return conductor. The 3rd harmonic is also
present on the phase/line because of the neutral line.

• If there is no 4th neutral wire connected (as in the diagram on the left), then the phases
do not contain any 3rd harmonic currents.

So what?

• In standard electric drive applications there is no neutral wire connected to the motor,
hence the 3rd harmonics are not present in the phases.

Page 56 of 100
fl
57. Power switch operating regions

• For high energy ef ciency, the power switches should be operated in the saturation
region.

• When operated in the active region, the power switches are not energy ef cient.

So what?

• The saturation region is the normal operating (conducting) state of a power switch. An
IGBT in the saturation region has a constant operating voltage of just under 2V.

• The active region is the transitional state of a power switch, during turn-on or turn-off.
Here, the voltage across an IGBT can be hundreds of Volts as it ramps up/down during
switching.

Page 57 of 100
fi
fi
58. Power switch losses

• Power switches have: switching losses, conduction losses and blocking losses (due to
the small current ow).

So what?

• Important to understand the mechanisms that create losses in inverters, as these losses
convert into heat and from a certain temperature level the inverter will start to limit the
power output to maintain that temperature (thermal derating).

• The blocking losses refer to the situation where the IGBT is in the OFF state, but positive
voltage is applied across it and the anti-parallel diode that prevents current ow. As the
diode blocks the current ow, it has some losses due to the reverse current ow (see
Note 60).

Page 58 of 100
fl
fl
fl
fl
59. Types of power switches

• Types of power switches (categories):

1. Uncontrolled (e.g. diodes)

2. Semi-controlled (e.g. thyristor - can turn it ON, can not directly turn OFF)

3. Fully-controlled - ON/OFF via gate control signal (e.g. IGBTs, MOSFETs)

So what?

• In modern applications one uses a combination of passive (uncontrolled) and active

(fully controlled) power switches to create the intended voltage and current waveforms
needed for the application.

Page 59 of 100
60. Diode turn-off

• Diode turn-off: there is a reverse current owing once the voltage across the diode
decreases below a certain level (typically 0.6-0.7V) - because the minority carriers
require a certain time to recombine.

• Minority carriers refer to electrons owing between the P and N junctions of the diode.

• T_recovery = reverse recovery time, as shown in the plot. During this time, there is a
very small leakage current owing in reverse through the diode.

So what?

• Useful to understand that diodes are not perfect on/off electronic devices. In reality,
there is a time associated with the diode turning off and there is also a small dynamic
leakage current associated with the turn-off transition.

• In power circuit design, engineers have to take this into account when selecting and
sizing the circuit components to avoid unintended turn-on or off of power switches that
should be off or on, respectively.

Page 60 of 100
fl
fl
fl
61. Diode turn-off recovery time

• Diode T_recovery can determine the maximum frequency of a fast switching device
(˜0.1µs).

• Diode I 2t rating is important (surge current)

So what?

• The recovery time of the anti-parallel diode in an IGBT for example, determines how fast
the IGBT can be turned back on again, after the previous turn-off: once the diode has
completed its turn-off transition, the IGBT can be turned on again - hence this waiting
time limits the maximum IGBT switching frequency.

• The I 2t rating refers to the ability of a diode to withstand large surges of current in short
time intervals - hence determining for a certain power level of the application, what is
the maximum frequency that the diode can sustain that power at.

Page 61 of 100
 62. Types of diodes

• Types of diodes:

1. Recti er diodes (slow but can work at up to 4500A)

2. Fast recovery diodes (in high frequency applications)

3. Schottky diodes (low voltage capability. Advantage: very low ON voltage drop)

4. SiC diodes (Silicon Carbide, ultra low loss, fast turn on/off=high switching frequency)

So what?

• Useful to understand the different types of diodes and which are suitable for what type
of application.

• In automotive applications, wide band-gap diodes are becoming more prevalent (SiC
and GaN - Gallium Nitride) due to the lower losses and higher switching frequencies of
up to 100kHz (as opposed to <10-15kHz for typical Si IGBT inverters).

Page 62 of 100
fi
63. Diode turn-off

• Schottky diodes: breakdown voltage approx. 100V, low current ˜300A

• SiC diodes: expensive, very fast, very reliable, ultra low ON-STATE losses

So what?

• Schottky diodes are not used in automotive electric drives because of their low voltage
capability (<100V - given that a HV battery voltage is in the range of 300-700V nominal).

• SiC devices typically offer a 2% increase in ef ciency of power inverters (roughly

96-98%), when used for both the active and passive switching elements in the power
module.

Page 63 of 100
fi
64. Control of transient voltage and current in power switches

• To control dv/dt during switch turn-on and turn-off, an RC (resistor- capacitor) network
(snubber) should be connected across the switch (see circuit drawing).

• R should be sized to limit the current during turn-ON

• To control di/dt, connect a small inductance in series with the device (as shown).

So what?

• Snubber RC circuits can be used in automotive inverters to control the rise and fall rates
of the voltage. Sizing of the R and C should take into account the full circuit - including
the source and the load.

• The control of the current rise/fall times is not frequent in automotive inverters, as an
inductor would be extra cost and packaging volume, also the motor windings have a
large inductance of their own and naturally prevent very large current slew rates.

Page 64 of 100
65. Turn-off snubber

• Turn off snubber - highlighted RC network.

So what?

• As described in Note 64.

Page 65 of 100
66. Transformer conceptual operation

• A transformer works like a differentiator - only passes the di/dt from the primary to the
secondary winding, unless the transformer becomes saturated

So what?

• The diagram shows a square pulse and how the transformer outputs the derivative of
this signal. Hence if the transformer is fed with a constant signal, the output will be null.

• Magnetic saturation was discussed in Notes 29,30,45. In this case, the output is also
saturated.

Page 66 of 100
67. Bipolar Junction Transistor (BJT)

• A BJT (bipolar junction transistor) should not be operated in the deep saturation area
because it is then dif cult to turn it off quickly, and also a danger of thermal runaway.

• It is recommended to operate the BJT in the quasi-saturation region by using an anti-

saturation network (Baker clamp)

• It is desirable to operate the transistor in the saturation region because the losses are
low (due to the low Vce in saturation). However, operation in the deep saturation region
is not feasible and operation in the quasi-saturation are is done. (See Note 57)

So what?

• Useful to understand that the control of power switches involves this consideration to
avoid the deep saturation region - even though the on-state voltage of the switch would
be lower (and hence the losses lower), it is to be avoided.

• The use of Baker clamps or Schottky diodes at the gate of the BJT is standard design.

Page 67 of 100
fi
68. Latching power switches

• Thyristor, TRIAC, GTO are latching device examples, only need a single pulse to turn on

• (In contrast) the BJT needs continuous base drive, requires continuous supply voltage.

So what?

• Interesting to understand this difference between latching power devices and fully
controlled power switches.

• For automotive electric drives, it is uncommon to nd latching power switches because

of the low switching frequency and limited control of turn-off - even though they do offer
this advantage that a continuous supply voltage for the control is not needed.

• See also Note 59.

Page 68 of 100
fi
69. BJT main problem

• Main problem of the BJT: very high drive current required (1-1.5A)

• Hence another transistor is used in cascade to feed the gate, called a Darlington
transistor.

So what?

• A Darlington transistor is basically a current ampli er type of transistor, with gains in the
order of 1000 times (input current to output current).

• This pairing of a BJT with Darlington works well, but is not typically included in modern
applications where ready-made power switches have been perfected to remove these
drawbacks.

Page 69 of 100
fi
70. Transistor overcurrent protection

• Transistors need adequate overcurrent protection. Fuses are too slow to protect an IGBT
or BJT.

So what?

• In automotive applications, overcurrent protection mechanisms of power switches are

implemented as standard using a fast short-circuit detection mechanisms that does not
involve the use of fuses (see Note 71).

• Overcurrent failure is one of the most common causes of inverter failure.

• Just for comparison, remember Note 54 - CSI switches are short-circuit proof.

Page 70 of 100
71. Transistor collector-emitter voltage (Vce)

• Vce is high when the transistor is in blocking mode, and low in conduction mode.

• In case of overload, Vce starts increasing during conduction. This is how the transistor
overcurrent is detected.

So what?

• A simple and effective short-circuit detection and protection mechanism, by monitoring

the collector-emitter voltage (Vce) during conduction. This is a standard method
implemented in all inverters with BJT/IGBT type transistors.

Page 71 of 100
72. BJT overload condition

• Monitor Vce of a BJT to detect overload condition. When Vce is high, the BJT is either
blocking or is in overload.

So what?

• Similar to Note 71, here the topic is current overload - which can also be the start of a
short-circuit transient. In both cases, the IGBT should be turned off for protection.

• The term blocking refers to the off state of the IGBT, where it creates an open circuit and
the voltage level at the collector is the HV battery DC voltage level (blocked at that
circuit point).

Page 72 of 100
73. BJT vs MOSFET

• BJT vs MOSFET, left & right hand side. G=Gate

So what?

• The BJT and the MOSFET carry out the same function - as power switches in an inverter/
converter type application. Also, both are controlled on/off via a Gate voltage.

• The differences are in the fundamental way that switching occurs, in the speed of
switching and hence the maximum possible switching frequencies, in the way losses
occur and in the capability in terms of peak voltage and current that they can handle.

• It is typical to nd MOSFETs in automotive applications today, especially of SiC, but not

BJTs (at least not in this classical form, but coupled with another type of switch).

Page 73 of 100
fi
74. MOSFET gate-source voltage

• The MOSFET is not cut off when the Vgs is smaller than a threshold. (Vgs=Gate-Source
voltage that controls the MOSFET on/off operation).

So what?

• The threshold voltage here refers to the constant speci c voltage of a MOSFET, which
sets the level for the gate-source voltage below which the MOSFET will turn off.

• However, just by controlling the Vgs to be below this threshold voltage is not enough to
ensure that the MOSFET is turned off.

• The speci cation of the threshold voltage is for a given drain current (D=Drain, See Note
73) and for a given temperature level (typical values are in the order of tens/hundreds of
uA and 25degC, respectively). If these reference values are exceeded, then even if the
Vgs is smaller than the threshold level, then the MOSFET will not turn off (or cut off).

• In practice if there is electrical noise in the form of capacitive coupling or common mode
currents circulating through the ground loop, this situation can easily occur - leading to
unintended ON states of the MOSFET and shoot-through in the inverter. It is not
uncommon for this type of fault to occur.

Page 74 of 100
fi
fi
75. BJT voltage drop

• The BJT (Bipolar Junction Transistor) has substantially lower voltage drop than the
MOSFET.

So what?

• The voltage drop refers to the on-state voltage of the BJT compared to the MOSFET.

• A low voltage drop is desirable over the power switch as this leads to higher ef ciency -
in terms of transferring more of the source voltage to the load, but also in terms of heat
build-up in the inverter.

• The higher the heat dissipation in the inverter, the more complex the cooling has to be -
adding cost, weight and volume to the unit.

• Note: the difference in voltage drop between the BJT and the MOSFET is current-
dependent - the higher the current, the higher the MOSFET voltage drop (as it works as
a resistor), whereas the BJT voltage drop is constant regardless of load current (within
the BJT operating range).

Page 75 of 100
fi
76. Paralleling BJTs

• Paralleling BJTs is dif cult because of their negative temperature coef cient, i.e. as the
current through the BJT increases it becomes less resistive. Hence the higher the
current, the lower its internal resistance R.

So what?

• When placing any circuit elements in parallel, the current is split on the basis of the
relative magnitude of the resistance values.

• Considering this, paralleling BJTs becomes dif cult because of the possible scenario
where an initial small resistance difference leads to the device with the lower resistance
receiving more current - this can lead to a cascading effect of increasing current which
can lead to the BJT over-current and irreversible damage. Additional circuit elements
are needed to control against this scenario - hence the dif culty in paralleling BJTs.

Page 76 of 100
fi
fi
fi
fi
77. Paralleling MOSFETs

• Paralleling MOSFTEs is OK because of their positive temperature coef cient.

So what?

• Following the same logic as explained at (76) but in the opposite direction, a scenario
where an initial small resistance imbalance occurs can only lead to a situation where the
current through the MOSFET drops to zero.

• Of course, this is not the intended operation but at least no further circuit design
considerations are needed to cater to the risk of over-current.

• Note: Why would one want to parallel MOSFETs and IGBTs? The reason is to increase
the current capability of the inverter, since individual power switches have a restricted
maximum current capability. In fact, paralleling power switches is done as standard by
power module manufacturers.

Page 77 of 100
fi
78. Transistor capacitances

• Transistor capacitances.

• Capacitance (internal) = Capacitance gate-drain (Cgd) + Capacitance gate-source (Cgs)

• The turn-on & turn-off times for the MOSFET are dependent on how fast this
capacitance is charged.

• The capacitance is in the order or picoFarads. Turn-on/off times are in the order of ns.

So what?

• This aspect is interesting for the understanding of what contributes to the dynamic
operation of transistors in general and in particular in this example for MOSFETs.

• These capacitances are the result of the structural design of the power switch, e.g. the
structure of the gate electrode or the p-n junction properties.

• Electronics engineers take into account these capacitances when designing the circuitry
around the power switches, together with any additional parasitic capacitances and
inductances, in order to avoid unintended turn on/off of the power switch. Depending
on the application, this can lead to inverter fault or complete failure.

Page 78 of 100
79. BJT vs MOSFET characteristics

• BJT vs MOSFET characteristics (Current vs. Voltage). Note that the axes should be
labelled in the opposite way - Voltage on the vertical axis and Current on the horizontal
axis.

• The MOSFET has both a linear (Resistive) region and a saturation region.

So what?

• The voltage shown in this plot is what is referred to in (75) as voltage drop.

• Interesting to see the conceptual difference between BJT and MOSFET. Note that actual
V-I characteristics can look very different depending on the voltage/current range of the
power switch.

• As the current (X-axis) increases, the BJT voltage (Vce) saturates quickly to a value that is
constant regardless of the current. The MOSFET has a linearly increasing voltage drop
(Vgs) with current, but also a saturation region at higher currents.

• As mentioned in (75) as low a voltage drop as possible is desirable. Hence for lower
current applications one may consider MOSFETs as the better option (the current below
the intersection point of the above curves).

Page 79 of 100
80. MOSFET drawback

• One of the drawbacks of a MOSFET is the higher voltage drop than the BJT.

So what?

• As described in (75, 79) the MOSFET has a higher voltage drop than the BJT which is not
desirable.

• However, it should be noted that the voltage drop is only related to the ON-state
conduction losses. The second main source of losses in power switches is the ON/OFF
or OFF/ON switching losses, where the MOSFET has lower losses than the BJT.

Page 80 of 100
81. MOSFET drawback

• MOSFET drawback: Cgd is varying with Vds (Drain-Source voltage).

• As Vds increases, Cgd decreases.

• See (78) for more information on MOSFET capacitances and nomenclature.

So what?

• As a general rule, circuit elements have a xed characteristic which the electronics
engineers use for their design and calculations.

• When any circuit element characteristic changes during operation as a function of

another variable, this makes the overall circuit design and calculations more complex.
Instead of a single operating point, a range of operating points have to be taken into
account which can have a cascading effect on the overall circuit.

Page 81 of 100
fi
82. BJT vs MOSFET comparison

• (i) BJT vs (ii) MOSFET comparison:

1. (i) Current-controlled vs (ii) Voltage-controlled

2. (i) Minority carrier device vs (ii) Majority carrier device

3. (i) Negative temperature coef cient vs (ii) Positive temperature coef cient

4. (i) Low ON-state power loss vs (ii) Higher ON-state loss

5. (i) Slower turn ON/OFF vs (ii) Fast ON/OFF

6. (i) Relatively higher power needed to drive it (1-1.5A) vs (ii) Low power required

So what?

• Many of these aspects have been described in detail in the previous notes.

• This is a good high-level overview of how the BJTs and MOSFETs compare, for general
understanding.

Page 82 of 100
fi
fi
83-84. BJT and IGBT

• A BJT has excellent ON-state characteristic

• IGBT = BJT + MOSFET

So what?

• The ON-state characteristic referred to here is the ON-state voltage drop of the BJT (see
also 75, 79)

• An IGBT is essentially a combination of BJT and MOSFET. The IGBT is currently the
standard type of power switch for high voltage electric drives in vehicles.

Page 83 of 100
- blank page -

Page 84 of 100
85. IGBT built-in diode

• IGBTs do not have a body diode, but they do have a built-in diode. This diode has the
role to optimise the turn ON/OFF performance.

So what?

• A body diode is intrinsic to the structure of the power switch - e.g. the MOSFET has a
body diode.

• This diode (either body or added/built-in) is connected in an anti-parallel way to the

switch - the purpose is to help block the forward current through the switch when it is in
the OFF state.

• In the case of IGBTs, the diode is added to reduce the time needed for the power switch
to turn ON or OFF. Even though a shorter ON/OFF time can potentially lead to higher
voltage spikes (higher order waveforms harmonics), the main advantage is lower
switching losses and the possibility to increase the switching frequency of the inverter
(leading to improved motor control and lower noise, vibration, harshness).

Page 85 of 100
86. ON-resistance of power switches

• The ON-resistance of the IGBT is lower than the ON-resistance of the MOSFET, hence
lower losses.

So what?

• Rds refers to the drain-source resistance value, this is the MOSFET notation. For IGBT the
equivalent would be the collector-emitter resistance.

• This statement is valid for high voltage medium-high power applications. For low
voltage applications (<60V) or below 20kW power for example, the MOSFET should be
considered also as a viable candidate (see note 79).

Page 86 of 100
87. IGBT turn-off current

• IGBT turn-OFF current

• The turn off current pro le includes an initial fast drop (MOSFET part of the IGBT) and a
subsequent slower decrease called Tail Current - due to the operation of the BJT part of
the IGBT (minority carriers).

So what?

• As described in previous notes, the BJT has excellent ON-state characteristics but the
main drawback of BJTs is the slow turn-off caused by the Tail Current.

• There are 2 main consequences of the Tail Current: 1. It limits the inverter switching
frequency (IGBT inverters are normally limited to 10-12kHz) and 2. The turn-off losses are
higher.

• Conversely, MOSFETs have worse ON-state characteristics but excellent turn ON/OFF
times. It is not uncommon for MOSFETs to have switching frequencies above 20kHz
(depends on the power of the load).

Page 87 of 100
fi
88-89. MOSFET input impedance

• MOSFET: very high input impedance —> very low current required to turn it ON
(remember that the MOSFET is voltage controlled)

• The high MOSFET input impedance is due to the two capacitances Cgs, Cgd (see #78)

So what?

• Also mentioned in note 82 (points 1 & 6), the MOSFET is voltage controlled and has low
turn ON/OFF power requirements.

• This note gives the reason for these attributes, i.e. the intrinsic capacitances of the
MOSFET (as described in Note 78) create a high input impedance.

• The input refers mainly to the gate-source connection where the applied voltage turns
the MOSFET ON (and its absence turns it OGFF).

• A low gate-source capacitance creates a high impedance to the driving voltage

(impedance is inverse proportional to capacitance). A high impedance for the same
applied voltage means a low current is needed to drive the MOSFET.

Page 88 of 100
- blank page -

Page 89 of 100
90. DC vs AC

• DC means average value of signal is FINITE

• AC means average value of signal is ZERO

So what?

• This is a basic question: how to de ne direct current (DC) and alternative current (AC)?

• Conceptually the topic is understood but for some applications a clear de nition is
needed - e.g. applications where both positive and negative currents are present, but
the average over time is not zero. Such an application is DC (e.g. switched reluctance
motors).

• The main types of traction motors (SM, IM) have currents and voltages with an average
of zero, hence we call them AC motors.

Page 90 of 100
fi
fi
91. Induction motor principle of operation

• Induction Motor (IM) principle of operation based on Lorentz law:

• Force = B x I x L, where B=rotating magnetic eld, I=current induced in the IM rotor

laminations, L = rotor conductor length.

So what?

• The drawing illustrates in principle how the Force is produced on the rotor of an
induction motor. The rotating magnetic led (induced into the rotor from the stator) in
combination with the owing current in the rotor bars creates a local force F which is
proportional to the length of the conductor in the magnetic eld.

• This local force F is constantly applied as the rotor is moving and creates the IM torque.

Page 91 of 100
fl
fi
fi
fi
92. Induction motor torque

• Induction Motor (IM) only produces torque when there is a difference in relative speed
between the stator magnetic eld and the rotor.

So what?

• This note points out the main characteristic of induction or asynchronous motors, which
is that the speed of rotation of the stator magnetic eld has to be different from the
rotational speed of the rotor.

• If the two speeds are the same, 0 torque is being produced.

• The difference in speed is also commonly referred to the Slip Angle (see note 93).

• Important to bear in mind that this is different from the Synchronous motor, where
torque is only produced if the two speeds are the same (synchronised).

Page 92 of 100
fi
fi
93. Induction motor slip angle

• IM - when slip = 0, the rotating magnetic eld does not cross through the rotor
conductors. Therefore, there is no current “induced” in the rotor laminations.

• Since I (current) = 0, then the Force F=BIL=0 (Lorentz Law, Note 91) and hence the
Torque is zero.

So what?

• The slip angle determines not just if torque is created, but also the polarity of the torque
(motor/generator mode) and the amplitude of the torque for the given voltage and
current to the motor.

• Normally each IM has its own xed characteristic of Torque vs Slip angle, which is
implemented in the motor control software.

Page 93 of 100
fi
fi
94. Induction motor operation

• The drawing illustrates the basic idea of a 2-pole IM rotor (represented by the circle)
and a 1-phase stator winding through which a current ia ows.

• When the current is positive, the MMF (Flux) is also positive (denoted as the alignment
North-South) and vice versa.

So what?

• The idea presented here is not something one would nd in practice, but it is a
fundamental concept to illustrate the basic principle of operation: if you have a 1-phase
motor with a 2-pole rotor, the rotor makes one rotation per one time period of the
current in the phase (here ia).

• In practice the motors are 3-phase (can also be more, for some applications - e.g. 6) and
the rotor has multiple magnetic poles.

Page 94 of 100
fi
fl
95-96. IM with 2 poles per phase

• In IM with 2 poles per phase: MMF dumbbell rotates 180deg in 1 current (ia) half cycles.

• In T, the MMF dumbbell makes 1 revolution for 1 pole pairs per phase.

• MMF speed (Magnetic ux speed, Ns) (RPM) = 60 x f, where f= current frequency in Hz

• In general, Ns = 60f/p , where p=no. pole pairs per phase, Ns=synchronous speed of
magnetic ux (stator).

Page 95 of 100
fl
fl
So what?

• The MMF dumbbell refers to the drawing of a 2-pole rotor (where the ends of the
dumbbell are the N and S magnetic poles)

• This note expands on the previous note and adds the general formula for synchronous
speed as a function of the frequency of the current (f) and the number of pole pairs per
phase.

• For example, if a motor has 8 rotor pole pairs per phase, for a current frequency of 50Hz
the rotor would spin in synchronism at 375 rpm.

• The formula for synchronous speed is only valid for synchronous motors, not for IM.

Page 96 of 100
97. Number of pole pairs per phase

• P = No. Pole pairs per phase. Each phase has its own number of pole pairs.

So what?

• The symbol p is the same as in Note 96. It is highlighted here that this is a per-phase unit
and not an absolute count of the magnetic poles of the rotor only. Typically electric
motors in electric drives have 3 phases.

Page 97 of 100
98. Induction motor - 2 pole pairs per phase example

• Diagrams showing a follow-up example from Note 94

• Here, the case is shown where the rotor has 4 poles (2 pole pairs) and 1 phase.

So what?

• By doubling the number of rotor magnetic poles (here 4, compared to 2 in Note 94), the
rotor spins half as fast for the same frequency of the stator current ia.

• In half the cycle for the current, the rotor rotates by one quarter of a rotation (90deg).
This is shown by the 2 diagrams at the bottom: left - rotor position at ia max(+), right -
rotor position at ia min (-).

Page 98 of 100
99. Induction motor = rotating transformer

• Think of an IM as a rotating transformer. When the stator frequency is equal to the rotor
frequency, it is like have DC (voltage) passed through a transformer: no current is
passed.

• Stator winding is equivalent to the primary (coil) of the transformer.

• Rotor aluminium bars are equivalent to the secondary (coil) of the transformer.

So what?

• An interesting analogy to grasp the key operating principle of induction.

• The stator frequency being equal to the rotor frequency is the same as the Slip Angle
being equal to zero (see Note 93).

• See Note 66 for the basic principle of transformer operation, as a connection to this
note.

Page 99 of 100
100. Two main types of induction motors

• Two main types of IM:

• (A) Cage rotor - with bars running axially and an end plate

• (B) Wound rotor - where instead of bars it has coil windings.

So what?

• These are just the rotor types - the stator can be the same for both.

• Each type of rotor has advantages and disadvantages in terms of losses, heating,
options for cooling, mass, ease of manufacturing.

• Which type of rotor is better depends on the speci cation of the application.

Page 100 of 100

fi