MEBS 6000 2010 Utilities services M.Sc.(Eng) in building ...
Transcript of MEBS 6000 2010 Utilities services M.Sc.(Eng) in building ...
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 1 of 38 July 2010
Introduction
Power electronics applies electronic techniques to the control of electric power. This
field is not new because vacuum tubes have long been used for this purpose. However,
more recently, power electronics has grown rapidly because solid state devices have
been developed, at decreasing price, to control electric power, specifically, power
transistors, silicon controlled rectifiers (SCR), and gate turnoff thyristors (GTO).
Switch
A switch is a device that has two stable states, ON and OFF. When ON, the switch has
an impedance much smaller than its load, and when OFF, it has an impedance much
larger than its load.
The concept depicted in the above diagram showing the relationship between current
through a switch against the voltage across it is important – semiconductors try to
simulate a mechanical switch using purely electronic technique. We will come back
on this.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 2 of 38 July 2010
Semiconductors
Semiconductors are neither good conductors nor good insulators. Semiconductors are
made from materials that have four valence electrons in their outer orbits.
Insulator Semi-conductor
[Adopted from Herman S.L. Industrial Motor Control]
Germanium and silicon are the most common semiconductor materials. Among the
two, silicon is used more often because of its ability to withstand heat.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 3 of 38 July 2010
Pure semiconductor material
[Adopted from Herman S.L. Industrial Motor Control]
When semiconductor materials are refined into a pure form, the molecules arrange
themselves into a crystal structure with a definite pattern. This is called a lattice
structure. To make semiconductor material useful, it is mixed with an impurity.
When pure semiconductor material is mixed with an impurity that has only three
valence electrons, such as indium or gallium, the lattice structure changes, leaving a
hole in the material
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 4 of 38 July 2010
P-type semiconductor material
[Adopted from Herman S.L. Industrial Motor Control]
This hole is caused by a missing electron. Since the material now lacks an electron, it
has a positive charge – this is called a p-type material.
When a semiconductor material is mixed with an impurity that has five valence
electrons, such as arsenic or antimony, the lattice structure has an excess of
electrons – the material has a negative charge. It is called an n-type material.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 5 of 38 July 2010
N-type semiconductor material
[Adopted from Herman S.L. Industrial Motor Control]
All solid state devices are made from combination of p- and n-type materials. The
number of layers and the thickness of each layer play an important part in determining
characteristics of the device formed.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 6 of 38 July 2010
Transistors
Transistors can be used as either amplifiers or electronic switches. In power
applications, transistors are used mainly as electronic switches. Two types of
transistors are commonly used in power applications: the bipolar transistor and the
field effect transistor (FET). Other power devices, such as the insulated gate bipolar
transistor (IGBT), are hybrids of these two types. The following figure shows various
sizes of bipolar power transistors.
Bipolar transistor
The bipolar transistor consists of three layers of semiconductor materials in the n-p-n
or p-n-p structure as shown in the following figure:
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 7 of 38 July 2010
The ratings of these transistors can be as high as a few hundred amperes. The
operation of the bipolar transistor is based on the capability of the p-n junction to
inject or collect minority carriers. When the emitter is forward biased, electrons are
injected from the n (emitter) to the p (base) region. If the other n layer (collector) is
reverse biased, the electrons in the p layer are collected in that n layer.
The base layer is very thin compared to the layers of the emitter or collector because it
is an obstacle to the flow of current. However, it serves a very useful purpose – it
controls the flow of electrons from emitter to collector. If a base current is injected in
the p junction, more current is allowed to pass from the emitter to the collector. The
relationship between the base and collector currents, however, is nonlinear.
The collector is positive biased with respect to the emitter or base, and the base
voltage is positive with respect to the emitter. The base-emitter junction is a simple
diode. Hence, the voltage difference between the base and emitter is very small (about
0.6V)
The basic equations of the bipolar transistor can be written as
CEOBC III += β
CBE III +=
BECBCE VVV +=
where β is the current gain (ratio of collector to base currents), and CEOI is the
leakage current of the collector-emitter junction. Because the leakage current is very
small compared with BIβ it is often ignored.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 8 of 38 July 2010
A transistor can be connected in a common-base or common-emitter form. The above
figure shows the characteristic of an n-p-n transistor connected in the common emitter.
The base characteristic is very similar to that of the diode. In the forward direction,
the base-emitter voltage is below 0.7V. A substantial increase in the base current
occurs at a slightly higher value of the base-emitter voltage.
The collector characteristics can be divided into three basic regions:
- the linear region,
- the cutoff region, and
- the saturation region.
In the linear region, the transistor operates as an amplifier, where β is almost
constant and in the order of a few hundreds. Any base current is amplified a few
hundred times in the collector circuit. This is the region in which most audio
amplifiers operate when using bipolar transistors. However, on a continuous basis,
power transistors do not operate in the linear region because losses of the transistor
are excessive in this region and can lead to thermal damage of the transistor.
The cutoff region is the area of the characteristic in which the base current is zero. In
this case, the collector current is negligibly small regardless of the value of the
collector-emitter voltage.
In the saturation region, the collector-emitter voltage is very small at high base
currents.
When used as a switch, a transistor operates in two regions only, viz, cutoff and
saturation. In the cutoff region, the transistor acts as an open switch, where the
collector current is almost zero regardless of CEV . In the saturation region, the
transistor operates as a closed switch because the voltage across the switch is very
small, and the external circuit determines the magnitude of the collector current.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 9 of 38 July 2010
The above figure explains the operation of a transistor in switching circuit. The
transistor is connected to an external circuit that consists of a dc source CCV and a
load resistance LR . The base circuit of the transistor is connected to a current source
to produce the base current of the transistor.
The loop equation of the collector circuit is represented by:
CLCECC IRVV +=
This equation, which demonstrates a linear relationship between CI and CEV , is
known as the load line equation. This load line equation has a negative slope and
intersects the CEV axis at a value equal to CCV and the CI axis at a value equal to
L
CC
R
V.
If the base current is set equal to zero, the operating point of the circuit is in the cutoff
region point 1. The collector current in this case is very small and can be ignored. The
collector emitter voltage of the transistor is almost equal to the source voltage CCV .
This operation resembles an open mechanical switch.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 10 of 38 July 2010
Now, assume that the base current is set to the maximum value. The transistor
operates in the saturation regions at point 2. The voltage drop across the
collector-emitter terminals of the transistor is small and can be ignored. The collector
(or load) current is almost equal to L
CC
R
V. In this case, the transistor is equivalent to a
closed mechanical switch.
The bipolar transistor is a current-driven device. To open the transistor, the base
current should be set to zero. To close the transistor, the base current should be set as
high as the ratings permit. Keep in mind that the base current must exist for as long as
the transistor is closed. Because β is small in the saturation region and the collector
current is high in power applications, the base current is also high in magnitude. This
situation creates two major problems: the first is that there are relatively high losses in
the base circuit. The second is that the driving circuit must be capable of producing a
large base current for as long as the transistor is closed. Such a circuit is large, of low
efficiency, and complex to build.
Even with the disadvantages of requiring high base current, thus requiring high power
base drive circuits and higher switching and operating losses than SCRs, power
bipolar transistor switches are gaining popularity especially in IGBT construction.
The main advantage is that turn ON and turn OFF are controlled by the base current,
and no forced-commutation circuit is required. Bipolar junction transistors do not
offer as high switching frequency as MOSFETs or IGBTs – high switching frequency
is very important in ac motor controller circuits. Therefore in LV (<200V)
installations, MOSFETs are more popular. However, BJT still plays an important role
in power electronics.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 11 of 38 July 2010
Example
A transistor has 1β =200 in the linear region and 2β =10 in the saturation region.
Calculate the base current when the collector current is equal to 10A, assuming that
the transistor operates in the linear region. Repeat the calculation for the saturation
region.
Answer
In the linear region
mAI
I CB 50
200
10
1
===β
In the saturation region,
mAI
I CB 1000
10
10
2
===β
Note that the base current in the saturation region is 20 times of that in the linear
region. This ratio is same as the ratio of 2
1β
β
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 12 of 38 July 2010
Field Effect Transistor (FET)
Field effect transistors (FETs) are widely used as electronic switches in computer and
logic circuits. There are several subspecies of FETs. The most common are the
junction gate FET (JFET), the metal oxide semiconductor FET (MOSFET), and the
insulated gate FET (IGFET).
The operation of FETs is based on the principle that the current near the surface of a
semiconductor material can be changed when an electric field is applied at the
surface.
In the above example, two n-junctions (source and drain) are embedded in a p
material. The gate, which is metal, is connected to the positive side of a dc supply.
The source and drain are connected to another dc supply, with the drain on the
positive side and the source on the negative side. The voltage difference between the
drain and source creates a current flowing in the channel. The magnitude of the
current is affected by strength of the electric field from the gate. Thus, the gate
voltage controls the drain current, which makes the FET much easier to control than
the bipolar transistor.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 13 of 38 July 2010
Several characteristics can be obtained from the many FET subspecies. The difference
between the two in the above two figures is that a MOSFET designed for
enhanced/depletion mode has a narrow-doped conducting layer diffused into the
channel. The presence of this layer results in current DI flowing inside the channel
even if the gate-to-source voltage GSV is negative.
The main advantage of FETs over bipolar transistors is in the way the current in the
switching circuit is controlled. The FETs are voltage-driven devices, unlike the
current-driven bipolar transistors. The gate voltage controls the drain current DI of a
FET, which is relatively easy to implement. For LV installations (<200V) MOSFET is
more popular than bipolar junction transistors.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 14 of 38 July 2010
Thyristors
Thyristor is a name given to a family of devices that include the silicon-controlled
rectifier (SCR), the bi-directional switch (triac), and the gate turnoff SCR (GTO).
These devices can handle large currents and are widely used in power applications.
Although not commonly used, other thyristor devices are also available for low
current control circuits, such as the silicon unilateral switch (SUS) and the bilateral
diode (diac).
Four level diode
The following shows the symbol and structure of a four level diode.
(Adopted from COGDELL, J.R. Foundations of Electric Power)
The structure consists of four alternating layers of p- and n-type semiconductor,
forming three pn junctions. 2 of the pn junctions face the same direction, and the
middle one faces the opposite direction. Thus one might anticipate that the device will
operate as three diodes in series, with the middle one turned around. Hence, no current
ought to flow in either direction, because at least one of the diodes will always be
reverse biased.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 15 of 38 July 2010
The characteristic displayed above reveals that no current flows for a negative
voltages, where 2 pn junctions are reverse biased, nor does current flow for positive
voltages, where the middle pn junction is reverse biased, until a threshold voltage or
breakdown voltage, VBO, is reached. After this threshold voltage is exceeded, the four
level diode begins to conduct freely: It “fires” acting as if the middle pn junction has
disappeared. The threshold phenomenon occurs because the doping levels in the fours
layers differ greatly. The outside p and n materials are doped heavily; hence they have
many carrier holes and electrons, respectively, available to diffuse into the middle n
and p regions, which are lightly doped only. Once the breakdown occurs in the middle
junction at VBO, the holes from above and the electrons from below flood into the
depletion region of the middle junction, where the electric field due to uncovered
charges reinforces their movement across the junction. Thus, this middle depletion
region effectively disappears due to the carriers from the forward biased junctions,
and we are left with two forward biased junctions in series. The small voltage for the
pnpn diode in the ON stage results from the contributions from each ON junction. The
turn ON voltage is typically less than 0.7V because the excess carriers from each
junction help each other.
If the current is reduced below a certain value called holding current Ih, the device
opens, and the current drops to zero. The voltage across the device is now equal to the
source voltage. This process is called commutation.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 16 of 38 July 2010
Thus, the pnpn has two states. It is OFF for negative voltage, and remains OFF for
positive voltage until the threshold voltage is reached, after which it turns ON. It
remains ON until the current is reduced to a small value, after which it turns OFF
again. The value of the threshold voltage, VBO, can be controlled over a modest range
by the semiconductor design. Typical pnpn diodes have threshold voltage from 6 to
32V. Thus, this device, like the pn junction diode, is a voltage controlled switch,
except that the four level diode requires much more than 0.7V to turn ON.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 17 of 38 July 2010
Silicon controlled rectifier (SCR)
The following figure shows the symbol and structure of a silicon-controlled rectifier
(SCR), which is sometimes simply called a thyristor.
(Adopted from COGDELL, J.R. Foundations of Electric Power)
The SCR has a pnpn structure with an external gate to turn ON the device. With no
gate current, the SCR characteristic is like that of a four level diode. The important
difference, however, is that the breakdown threshold occurs at a much higher voltage,
indeed, high enough that the SCR should never conduct because the input voltage will
never exceed its threshold. On the contrary, the forward biased SCR should fire only
when a pulse of current is delivered to the gate. This is shown by the IG>0
characteristic in the following figure; in effect, the threshold is reduced to a very small
value when the gate conducts.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 18 of 38 July 2010
Physically, the gate injects holes into the lightly doped p region and floods the
depletion region with carriers, thus initiating breakdown.
The turn on voltage, VTO, is dependent on the magnitude of the gate current – the
higher the gate current, the lower the turn on voltage. When the gate pulse is as high
as the rating permits, the SCR can be turned on at a very low anode-to-cathode
voltage. A general expression relating the turn on voltage VTO to the equivalent dc
gate current IG can be written as
KIBOTO
GeVV −=
where K is a constant whose value is dependent on the device characteristics. Note
that this equation is purely empirical, and specification of the particular device should
be consulted for accurate information about SCR triggering characteristics.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 19 of 38 July 2010
The popularity of the SCR is due to several factors, including the following:
- SCRs are cheaper to manufacture than other types of solid state switches, such as
the bipolar transistors and FETs
- A single pulse, instead of the continuous signal needed by a bipolar transistor,
can turn on a SCR. Hence, losses are reduced.
- In ac circuits, the SCR is self-commutated and may not need an external circuit to
turn it off.
- As SCR can have much larger current and voltage ratings than the transistor.
SCRs are usually found in large rating power electronic circuits, including large rating
circuits, but is becoming less popular in smaller variable frequency drives.
In some literature and in commercial applications, the terms thyristor and SCR are
frequently used interchangeably. Currently, SCRs are frequently used in power
electronic circuits. However, this may change in future as other devices such as the
IGBT are getting larger ratings and are easier to control.
Commutation
Commutation refers to the switching of a conducting device from the ON state to the
OFF state. An SCR must be commutated by reducing its current below the holding
current value required to sustain conduction. Normally, this is accomplished by
reverse biasing the device for a period of time. When in an ac cycle, the voltage
across an SCR changes from forward bias to reverse bias, the SCR ceases to pass
current and is line commutated.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 20 of 38 July 2010
When an auxiliary circuit is used to commutate the SCR independent of the line
voltage, the SCR has a forced, or device, commutation. A lot of efforts has be devoted
to design circuits for device commutation of SCR to ensure safe operation.
When the device is commutated by controlling the gate signal, it is self-commutated.
SCR cannot be self commutated and must be line commutated. On the other hand,
transistors has low reverse blocking ability, thus cannot be line commutated and must
be self commutated.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 21 of 38 July 2010
Example
A SCR is connected in series with an ac voltage source of 120V (rms value) and a
load resistance. The breakover voltage of the SCR VBO=200V, and K=0.2mA-1.
Calculate the approximate value of the dc gate current required to trigger the SCR at
30o.
Answer
The source voltage can be written as
( ) ( )tVs ωsin1202=
When the SCR is open, the voltage across the SCR is equal to the source voltage. For
a 30o triggering angle the voltage across the SCR is
( ) ( )30sin1202=sV
The dc triggering current can then be calculated as
KIBOTO
GeVV −=
( ) ( ) ( )2.020030sin1202 GIe−=⇒
mAIG 29.4=⇒
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 22 of 38 July 2010
Gate turnoff thyristors (GTO)
GTOs are similar to SCRs except that conduction through the GTO can be stopped by
applying a negative voltage – negative with respect to the cathode – to the gate.
(Adopted from COGDELL, J.R. Foundations of Electric Power)
The above figure shows that the circuit symbol for a GTO is like a SCR symbol
except for a mark on the gate. The forward biased GTO is turned ON by a pulse of
positive current to its gate, but unlike the SCR, it is turned OFF by a pulse of negative
gate current. The negative current removes carriers from the cathode region, and the
inner pn junction blocks the forward current.
GTOs are thyristors able to handle a greater amount of current than transistors and are
used in large power rating power electronic circuits.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 23 of 38 July 2010
Diacs and triacs
The following diagram shows a non-rectifying power controller.
(Adopted from COGDELL, J.R. Foundations of Electric Power)
Diac is the short form for diode for alternating current. In the above diagram the
device replacing the four level diode is called a diac. It is like two parallel four level
diodes facing in opposite direction and fires in either direction. Most diac have a
breakdown voltage of about 30V.
[Adopted from HERMAN, Stephen L. Industrial Motor Control]
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 24 of 38 July 2010
[Adopted from HERMAN, Stephen L. Industrial Motor Control]
Similarly, triac is the short form for triode for alternating current. In the above wiring
schematics, the device replacing the SCRs is called a triac, and it functions like two
parallel SCRs facing opposite directions, but with their gates joined together. Once
triggered, a triac continues to conduct until the current through it drops below the
threshold value, i.e. the holding current, at the end of a half cycle of an ac supply. This
makes the triac a very convenient switch allowing the control of very large power
flows with only small control current typically in the order of milliampere only.
The following figure shows how the load voltage varies.
(Adopted from COGDELL, J.R. Foundations of Electric Power)
This is the preferred circuit for light dimmers and universal motor tools, which do not
require dc voltage.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 25 of 38 July 2010
Darlington transistor
When a bipolar transistor operates as a switch, only the cutoff and saturation regions
are used. In the saturation region, the current gain β is very small. Hence, when the
transistor is closed, a large base current is needed. This base current must be
maintained for as long as the transistor is closed. The continuous large base current
results in high transistor losses and demands an extensive control circuit to provide.
For example:
Load current (emitter current) is 100A, β is 4, the base current must then be
( )β+1
100=20A
This base current is very large.
To reduce the base current, two transistors can be connected in Darlington fashion.
The emitter current of 1Q is ( ) 111 BIβ+ .
The emitter current of 2Q is ( ) 221 BIβ+ = ( )( ) 112 11 BIββ ++
Hence, the ratio of the emitter current of 2Q (which is the load current) and the base
current of 1Q (which is the triggering current of this Darlington transistor) is
( )( )121
2 11 ββ ++=B
E
I
I
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 26 of 38 July 2010
For example:
1Q and 2Q are identical transistors with 1β = 2β =4. If the load current is also 100A,
then the base current for the Darlington transistor is ( )( )12 11
100
ββ ++=4A
This is only one fifth of the base current computed above for the single transistor.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 27 of 38 July 2010
IGBT (insulated gate bipolar transistor)
Bipolar transistors are current controlled devices with relatively low losses in the
power circuit (collector circuit) during the conduction period, due to their relatively
low forward drop CEV when closed. It is suitable for current ratings up to a few
hundred amperes. Bipolar transistors are more suitable for high switching frequencies
than SCRs. These are very desirable features for power applications. However, bipolar
transistors have very low current gains at the saturation region (when closed). Thus,
the base currents are relatively high, which makes the triggering circuits bulky,
expensive, and of low efficiency.
On the other hand, MOSFETs are voltage-controlled devices that require very small
input current. Consequently, the triggering circuit is much simpler and less expensive
to build. In addition, the forward voltage drop of DSV of a MOSFET is small for low
voltage devices (<200V). At this voltage level, the MOSFET is a fast-switching power
device. Because of these features, MOSFETs replace bipolar transistors in
low-voltage applications.
In high-voltage applications (>200V), both the bipolar transistor and the MOSFET
have desirable features and drawbacks. Combining the two in one circuit, as shown in
the following figure, enhances the desirable features and diminishes the drawbacks.
[Adopted from EL-SHARKAWI, Mohamed A.,
Fundamentals of Electric Drives.]
[Adopted from Murphy & Turnbull, Power Electronic Control of
AC Motors]
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 28 of 38 July 2010
The MOSFET is placed in the input circuit and the bipolar transistor in the output
(power) circuit. The MOSFET is triggered by a voltage signal with a very low gate
current. Then the source current of the MOSFET triggers (closes) the bipolar
transistor. The losses of the output power circuit are relatively low even for
high-voltage applications. Furthermore, because the output circuit is a bipolar
transistor, it can be used in higher frequency switching applications (when compared
with thyristors). These two devices can now be included on the same wafer; the new
device is called the insulated gate bipolar transistor or IGBT.
As the name implies, IGBT has an “insulated” gate, i.e. very high impedance, so
IGBT is a voltage, not current, controlled device.
Frequency inverters using IGBT on the dc/ac inverter section does not use SCR in the
ac/dc conversion section, thus subject to less problem of line harmonics. This
self–commutating device is available up to 300-Ampere current ratings, has good
turn-on and turn-off ability (its voltage control feature needs only 3 to 5 Volts of
energy to turn on) and has switching speeds of up to 18 kHz. It is cost effective to
manufacture and can be implemented into an electric circuit at relatively low costs.
All these features make the IGBTs gain popularity in VFDs today.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 29 of 38 July 2010
With costs and performance driving the semiconductor industry, more efficient
versions of IGBTs are coming out very year, and VFD designs change very often.
The following diagram depicts the general performance limits and application range
of different power electronic devices.
(Adopted from BARNES, M. Variable speed drives and power electronics)
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 30 of 38 July 2010
Device Current ON/OFF Ideal switch Bidirectional Instantaneous turn on / turn off 0 on state impedance, infinite off
state impedance Diode Unidirectional Forward voltage turn on / reverse voltage
turn off
SCR Unidirectional Turned on by a +ve gate pulse, cannot be self commutated
Forward and reverse blocking ability, thus can be line commutated
In absence of gate pulse, can also be turned on at high anode to
cathode voltage, or high dt
dv
GTO Unidirectional Turned on by +ve gate pulse, can be turned off by negative pulse at gate or by line commutation
Forward and reverse blocking ability, thus can be line commutated
A variant, asymmetric GTO, has low reverse blocking voltage
Triac Bidirectional Turn on: +ve or –ve gate pulse associated with corresponding forward or reverse voltage. Turn off: line commutation
Symmetrical forward & reverse blocking. Ideally suited to phase angle firing. Low voltage, low power, low frequency (<400Hz)
BJT Unidirectional +ve gate current to turn on, removal of gate current to turn off
Low reverse blocking ability, need to be self commutated
MOSFET Unidirectional +ve gate voltage to turn on, removal of gate voltage to turn off
Low reverse blocking ability, need to be self commutated
Very fast switching frequency
IGBT Unidirectional +ve gate voltage to turn on, removal of gate voltage to turn off
Low reverse blocking ability, need to be self commutated
Low on state losses, high switching frequency
[Table adopted from SHEPHERD, W., ZHANG, L. Power converter circuits.]
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 31 of 38 July 2010
Snubbing circuit
To protect a power electronic switch against excessive dt
dv and
dt
di, a snubbing
circuit must be used. From the instant when voltage is applied to a switch, there is a small delay before the
switch turns ON, and there is a rate of rise of the switch current dt
di, which can cause
device failure due to localized heating in the junction. If it is not limited by load inductance, an external inductor must be added in series with the switch to limit the
dt
di of the gate.
Also critical is the rate of voltage rise of the switch in its OFF state. In certain applications, a rapid voltage increase can initiate conduction, independent of the gate
signal, and cause device malfunction. The snubber circuit limits the dt
dv across the
gate. With the switch OFF, the small (10 to 100Ω ) resistor in series with the capacitor
and the inductance in the load circuit limit the dt
dv across the switch, and the
capacitor blocks dc current. When the switch fires, the energy stored in the capacitor is dissipated in the resistor and the switch.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 32 of 38 July 2010
Let us first assume that the load has the following impedance:
LLLL Cj
LjRZω
ω 1++=
Now look at the path of the current 1i , which can be written in the Laplace form as
( ) ( )
++=
sCsLR
sVsI
11
where sL RRR +=
sL LLL +=
sL CC
C11
1
+=
and the subscripts L and s refer to the load and the snubber respectively. Suppose that there is a step input of source voltage, hence
( )s
VsV =
where V is the magnitude of the step change of voltage. Thus
++=
sCsLR
sVI
1/
1
++=
LCs
L
Rs
LVI
1/
21
Now let natural frequency of oscillation be
LCn
1=ω (note that this is not synchronous speed of motor)
and damping coefficient be
L
CR
2=ζ
So that
21nC
Lω=
nL
R ζω2= , and
21nLC
ω=
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 33 of 38 July 2010
The s-domain equation can thus be re-written as
[ ]22
2
12 nn
n
ss
VCI
ωζωω
++=
The inverse Laplace transform of this equation will give the time-domain function of the current 1i (inverse Laplace transform can be found by looking up table):
( )[ ] teVC
ntn n 2
21sin
1ζω
ζω ζω −
−−
Differentiating this equation with respect to t will give
( )[ ] ( )[ ] teVCteVC
dt
din
tnn
tn nn 222
2
2
1cos1sin1
ζωωζωζ
ζω ζωζω −+−−
−= −−
Let us assume that the capacitors are initially uncharged. Furthermore, the charge on
the capacitors cannot instantly change. The maximum dt
di then occurs at the initial
time (t=0). By substituting t=0 into the above equation, hence
L
V
dt
di =max
The snubbing inductor is then calculated as
L
mas
s L
dt
di
VL −=
For adequate protection, sL should be selected so that it can V will not exceed the
breakover voltage, BOV of the device, and dt
di will not exceed, say, half the
maximum dt
di rating of the device. Therefore,
L
rated
s L
dt
di
VL −
=
5.0
This should be adequate in protecting the device from excessive dt
di due to supply
surges. This is only one of the three transients that a device should be able to
withstand without damage. The other two are the dt
dv and
dt
di created by the RC
snubbing circuit itself when the device is turn ON.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 34 of 38 July 2010
The RC circuit ( sR and sC ) can protect the device from the other two transients. Let
us assume that the charge on the capacitor sC is zero when the voltage V is applied.
With this assumption, the voltage across the device SCRV at the initial time is
1iRV sSCR = Then
dt
diR
dt
dVs
SCR 1=
But now L
V
dt
di =max
at time t=0, so
L
VR
dt
dVs
SCR =
This shows that the smaller the resistance sR , the smaller the dt
dv across the device.
However, sR is needed to limit the dt
di created by the snubbing capacitor.
Let us assume that the device is triggered. The current going through the device has two components, one is 1i and the other is 2i from the snubbing capacitor. We have
already discussed the dt
di attributed to 1i . The current 2i will also cause a
dt
di and
must also be limited to a tolerable value. Now
( )ssCRt
s
eR
Vi /02
−=(see Appendix 1)
where 0V is the capacitor voltage due to its initial charge before the device is
triggered. We may assume the worst case value for 0V as BOV . The dt
di of the
circuit is then
ssCRt
ss
eCR
V
dt
di −−= 202
The maximum dt
di2 occurs at time t=0. Hence
ss CR
V
dt
di2
0
max
2 −=
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 35 of 38 July 2010
As in the previous case for dt
di1 , dt
di2 should also be limited to say, half of the
device’s rating.
Now if sR is small, dt
dv is small but then
dt
di2 will be large, a compromise shall be
made.
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 36 of 38 July 2010
Example A SCR is connected between an ac source and a resistive load. The maximum allowable current transient of the SCR is 100sA µ , the maximum non-repetitive forward blocking voltage is 300V. Maximum allowable voltage transient of the SCR is 1500 sV µ . Calculate the value of the resistance, inductance and capacitance of the snubbing circuit to protect the SCR from excessive voltage transient and limit the current transient to half of the maximum rating. Answer
L
rated
s L
dt
di
VL −
=
5.0
As the load is resistive, LL =0, so
( )( )6101005.0
300
×=sL
hLs µ6≥ Now
L
VR
dt
dVs
SCR =
66
106
300101500 −×
=× sR
Ω≤ 30sR Also
ss CR
V
dt
di2
0
max
2 −=
( )( )sC2
6
30
300101005.0 =×
FCs µ0067.0≥ To reduce the size of the inductor, iron core material could be used. The problem with iron core inductors, however, is the core saturation, which reduces the values of the inductance at high current values. Air core inductors do not suffer from saturation but are bulky. Nevertheless, air core inductors are normally used for snubbing circuits. A sR of Ω30 will incur excessive power loss in the switch, we would manipulate
within the allowable figures. Let’s choose hLs µ10=
sAL
V
dt
di µ301010
3006
1 =×
== −
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 37 of 38 July 2010
This is less than the rating of 100 sA µ , thus acceptable. If we choose, sC as Fµ1
ss CR
V
dt
di2
0
max
2 −=
( )( )( )66 101101005.0
300−××
=⇒ sR
Ω=⇒ 45.2sR Then
sVL
VR
dt
dVs
SCR µ5.731010
30045.2
6=
×== −
This is less than the rating of 1500 sV µ thus acceptable. Another factor that should be considered is the losses of the snubbing circuit due to the presence of sR . When the SCR is not triggering, the current 1i causes losses in
the snubbing circuit in the form of sRi 21 . A bypass diode can be used in parallel with
sR to reduce the losses. This diode will also reduce the voltage transient. [Text and figures mostly adopted from SHEPHERD, W., ZHANG, L. Power converter circuits, and EL-SHARKAWI, Mohamed A., Fundamentals of Electric Drives.]
MEBS 6000 2010 Utilities services M.Sc.(Eng) in building services
Faculty of Engineering University of Hong Kong
K.F. Chan (Mr.) Page 38 of 38 July 2010
Appendix 1 Voltages across the resistor, capacitor and the switch can be written as:
( ) ( ) ( )∫+= dttiC
tiRtvs
s 220
1
Taking Laplace transform
( ) ( ) ( )s
s sC
sIsIRsV 2
2 +=
( )
ss sC
R
sVI
10
2
+=
Now suppose that there is a step input of voltage of magnitude 0V , hence
( )s
VsV 0
0 =
Therefore
ss sC
R
sVI
10
2
+=
sCR
VCI
ss
s
+=
10
2
Taking inverse Laplace transform
( ) ( )ssCRt
ss
s eCR
VCti /0
2−=
( ) ( )ssCRt
s
eR
Vti /0
2−=⇒