Variable Frequency Controllers
The speed of standard induction motors can be controlled by
variation of the frequency of the voltage applied to the motor.
Due to flux saturation problems with induction motors, the
voltage applied to the motor must alter with the frequency.
The induction motor is a pseudo synchronous machine and so
behaves as a speed source. The running speed is set by the
frequency applied to it and is independent of load torque
provided the motor is not over loaded.
Modern Variable Frequency drives come in two major formats,
V/Hz and vector. The V/Hz drive is a drive where the voltage
applied to the motor is directly related to the frequency.
In the idea motor, the magnetic circuit would be purely inductive
and keeping a constant V/Hz ratio would maintain a constant
flux in the iron. The real motor has resistance in series
with the magnetising inductance. This has no bearing on the
operation at line frequency, however as the frequency of the
drive is reduced, the resistance begins to become significant
relative to the inductive reactance. This causes the flux
to reduce at very low frequencies and so it is difficult to
get sufficient torque at low speeds. For many applications,
this low torque is not a problem, but there are some that
do need a high torque from a low speed. Early drives were
designed with a voltage boost to provide a measure of torque
increase at low speed.
Vector drives have a mathematical model of the drive in software
and by measuring the current vectors in relation to the applied
voltage, they are able to maintain a constant field at all
frequencies below the line frequency. These drives need to
be tuned to the motor and typically include a self tuning
algorithm that is enabled at commissioning to determine the
component values for the mathematical model. If the motor
is replaced, the drive needs to be retuned to learn the characteristics
of the new motors.
Vector drives come in three major formats, closed loop, open
loop and direct torque control. The closed loop controllers
were the first vector controllers and are still the best option
for accurate control at zero speed. The open loop vector and
DTC are suitable for applications requiring good control above
3 – 5 Hz.
Quite a number of modern drives can operate as V/Hz, open
loop vector or closed loop vector just by changing a parameter.
– closed loop requires a shaft encoder to give accurate speed
feedback.
The major differentiation between modern VSDs are the enclosure,
auxiliary functionality, programming and user interface. Low
cost drives are often very poorly filtered and can create
major RFI (EMC) issues. Some drives include no filtering and
must be installed with external filters, and others include
all the filtering required.
DC and AC reactors help to reduce the noise generated by the
drive, and to improve the distortion power factor of the drive.
Because the drive rectifies the incoming supply, the current
waveform is very distorted and so the harmonics are high.
Low cost drives without the reactors have a very poor power
factor. NB Most drive manufacturers quote the COS (phi) as
better than 0.95 implying a high power factor. While the displacement
power factor is high, the distortion power factor can be less
than 0.7 Distortion power factor can not be corrected with
capacitors, but can be improved with expensive filters. There
are “active front end” drives or “regenerative” drives that
have an inverter stage on the input as well as the output
and these can draw sinusoidal current from the supply resulting
in a high power factor. It is possible that this technology
may become a mandatory requirement at some time in the future.
Drives are typically used in some form of automation process
and so they are now including additional functionality and
controls to simplify the automation process. There are a number
of programmable inputs and outputs and relays and most drives
also include a PID loop and a motorised pot is also common.
PID information.
Vector drives and some V/Hz drives can be set up for speed
control or torque control. Torque control is used in tensioning
applications such as paper machines where the master controls
a winding drum and the diameter increases as the drum fills
up. This requires other drive feeding the paper to run at
different speeds. Traditionally, this was achieved by DC machines
as they naturally operate in torque mode.
Design
The VSD power sections comprise an AC rectifier to convert
the incomming power from AC to DC. This is followed by a power
DC Filter which comprises a number of high voltage high current
DC capacitors commonly in a series parallel arrangment. The
DC filter will commonly include one or two DC chokes in series
with the rectified DC.
After the DC Filter, comes the Output inverter stage which
is made up of a series of solid state switches. There are
three arms for a three phase output with two switches on each
arm. One switch connects the positive DC bus to the output
of that phase, and the other switch connects the negative
DC bus to the ouput on that phase. Control of the output switches
produces a PWM output waveform designed to cause a sinusoidal
current to flow into the motor. There are a number of schemes
and algorithms for the generation of the output waveforms,
one common algorithm is the space vector modulation technique.
The waveform generation is usually done in firmware or in
a special function chip.
.
AC to DC Converter
The AC to DC converter is a full wave bridge rectifier, single
phase or three phase depending on the input requirements.
The rectifier can be controlled using a combination of SCRs
and Rectifiers, or more commonly uncontrolled using rectifiers
only. Because the output of the rectifier is connected to
a large capacitive filter, there must be a means of providing
the initial charge to the capacitors without damaging the
rectifier. - The initial charging current for discharged capacitors
connected to the full rectified voltage is very high and would
cause rectifier failure.
The initial charge current is commonly limited by a series
resistance in one of the DC outputs. This soft charge resistance
is shorted out as soon as the capacitors are fully charged.
The shorting device can be a relay or contactor, or it can
be an SCR. The alternative means of limiting the charge current
is to use a controlled bridge and slowly increase the output
voltage applied to the filter.
DC Filter
The DC filter provides smoothing of the DC bus applied to
the output DC to AC inverter. There must be sufficient capacitance
to provide the smoothing required for the output current required.
The capacitors must have sufficent ripple current rating to
avoid excess heating and life shortening and voltage rating
to withstand the maximum expected input voltages. There are
two types of DC filter used, a capacitive input filter and
an inductive input filter. The capacitive input filter comprises
a capacitor bank and an inductive input filter has an inductor
in series with at least one of the DC inputs to the capacitive
filter.
With the capacitive input filter, current will flow from the
supply, through the rectifiers into the capacitors only when
the supply voltage is higher than the DC voltage. The result
of this is that a very high current flows for a short time
at the crest of the waveform only. This results in a very
low distortion power factor, lot of harmonics and excessive
heating of the rectifier and capacitors. The addition of the
DC Bus Choke(s) is that a lower current flows for longer in
each half cycle reducing the harmonics and increasing the
distortion power vactor. Another advantage of the DC Bus choke
is that it helps to decrease the amount of switching noise
that leaks back on to the supply, reducing EMC radiation.
The filter values are very different for single phase inputs
and three phase inputs due to the magnitudes and frequency
of the ripple currents. For a single phase input, the ripple
frequency is twice line frequency and for a three phase input,
the ripple frequency is six times the line frequency.
DC to AC Output Inverter
The AC output inverter for a three phase output stage comprises
six solid state switches. In small low voltage and low current
VSDs, the output stages will typically be MOS FETs and in
larger VSDs, they are typically IGBTs.
The output switches operate at a high frequency, typically
between 3 KHz and 16KHz, and are controlled to produce a PWM
output waveform which causes a sinusoidal current to flow
in the motor. There are many different pwm schemes and algorithms
with different advantages. One common waveform generator scheme
is the Space Vector Modulation algorithm. SVM is covered here.
The output voltage must provide both variable voltage and
variable frequency control.
Each switching element needs
to have a driver circuit that is isolated from the control
electronics and is able to provide sufficient energy to fully
control the switching elements. In some cases, this would
mean three isolated supplies to run the three top switching
elements, and one isolated supply to run the bottom switching
elements. The circuitry must be capable of withstanding very
high rates of change of voltage with minimum delays. Care
must be taken to prevent the upper and lower switch on one
phase being on at the same time, this includes through the
switching stage. This requires an interlock delay between
one switch turning OFF and the other switch turning ON.
Braking
Rapid slowing of the load can require energy to be removed
from the load. This energy goes back into the drive and will
result in an increasing DC bus voltage. If the bus voltage
goes too high, the drive will be damaged.
The excess energy can be dumped out into large resistors provided
that the drive is fitted with a braking module, or can be
fed back into the supply if the drive has an active front
end. If there are multiple drives in operation but with different
duty cycles, it is possible to common all the DC bus circuits
and the excess energy can then go into driving other motors.
The Braking resistors need to be sized to suit the drive (resistance)
and to suit the load (Brake energy).
Since adjustable frequency
controllers typically accelerate a motor and load by slewing
the motor voltage and frequency in such a way as to remain
in a region of operation above "breakdown RPM" (as
illustrated in Figure 1), the usual constraints of fixed voltage,
fixed frequency starting and acceleration do not apply. Starting
torque and current are no longer functions of the 1.0 per
unit slip characteristics of the motor but are limited by
the overload capability of the control. Thus, the controller
can be matched to the motor in such a manner as to produce
the appropriate starting torque based on a torque/amp ratio
equal to that under full load conditions. By evaluating the
drive as a motor and control "package", the motor
designer can take advantage of this to enhance the level of
starting torque as well as overload torque per amp as shown
in Figure 2.
Figure 1
Fixed Voltage and Frequency Speed Torque Curve
Figure 2 
Overload Torque Per Amp
--------------------------------------------------------------------------------
Peak Currents
In addition to the RMS current
level, an important rating point for a transistor (typically
used in adjustable frequency controllers) is the peak current
capability. The high frequency transient current which results
from the electronic switching of the control output voltage
is inversely proportional to the leakage inductance of the
motor. As noted in Table 1 the leakage inductances can be
increased by altering the design of the windings and the magnetic
cores in the motor. The use of an electromagnetic design specifically
for adjustable frequency power can significantly reduce the
peak current required for a given level of power output (see
Figure 3). This will not only improve the reliability of the
drive, but often can prevent costly over sizing of the AC
controller and provide the most cost effective solution.
Adjustable-Frequency
Definite-Purpose Design Standard Motor Design
Figure 3 
Typical PWM Current Waveforms
Definite-purpose, adjustable
frequency design reduces peak as well as RMS current required
from the controller for a given horsepower.
--------------------------------------------------------------------------------
Motor Heating
One of the more obvious sources
of increased stress on an induction motor insulation system
is higher operating temperature when run on variable frequency
controllers. The higher operating temperatures are the result
of increased motor losses and often reduced heat transfer
as well. As a result, many standard efficient, fixed frequency
design motors will not achieve their nameplate rating when
operated on an adjustable frequency control at 60 Hz while
remaining within temperature limits. While these elevated
temperatures may not lead to an immediate insulation failure
they will result in a significantly shorter life. In most
modern insulation systems, a 10 degree Celsius increase in
operating temperature will result in a 50% reduction in expected
life. This is one of the reasons why "High Efficient"
designs, which have inherently greater thermal reserves, are
often recommended for operation on adjustable frequency controls.
When an induction motor is
run with voltage and current waveforms as seen in Figures
4a through 4d, the deviation from the ideal sinusodial waveshapes
create additional losses without contributing to steady state
torque production. The higher frequency components in the
voltage waveform do not increase the fundamental air gap flux
rotating at synchronous speed. They do, however, create secondary
"hysteresis loops" in the magnetic steel, which
along with high frequency eddy currents produce additional
core losses and raise the effective saturation level in the
lamination material. As another consequence of these higher
frequency flux variations there are higher frequency currents
induced in the rotor bars which generate additional losses.
Appropriate electromagnetic design, including rotor bar shape
can minimize these added losses.
The higher frequency components
of the current waveform also do not contribute to the steady
state torque. They do, however, increase the total RMS current
resulting in added I R losses in the stator winding. In addition
to higher frequency current components there can also be low
frequency "instabilities" in the currents seen by
the AC motors on variable frequency controllers. These asynchronous
components of current again cause added losses without contributing
to the steady state torque production. Motor designs which
help minimize harmonic currents lead to lower I R losses.
--------------------------------------------------------------------------------
Typical Waveforms from Adjustable
Frequency Controllers
Figure 4A 
Voltage at 50% of Base Speed
Figure 4B
Current at 50% of Base Speed
Figure 4C 
Voltage Near Base Speed
Figure 4D 
Current Near Base Speed
--------------------------------------------------------------------------------
Motor Cooling
As has been well documented
in the literature, when AC motors are run across a wide speed
range their heat transfer effectiveness will vary a great
deal. Cooling fans whose rotation is directly supplied by
the motor are subject to high windage losses and noise at
high speeds. Modern AC controllers are capable of operating
across a very wide frequency range, often up to several hundred
hertz. While this provides great flexibility in the control,
it places the motor cooling fan well above its fixed frequency
design operating point which often leads to inefficient air
flow and objectionable noise. In low speed operation the fan's
effectiveness falls off with the motor's speed. Figure 5 shows
typical cooling curves for a family of totally enclosed fan
cooled motors. In variable torque applications this reduction
in cooling air often stays in balance with the reduction in
motor losses as the load is reduced with speed. However, in
constant torque applications the motor's temperature limits
will likely be exceeded. An independently powered blower can
provide an essentially constant heat transfer rate. Although
not a standard fixed frequency motor feature, depending on
the load/speed profile required by the application, this can
be a very effective choice and is often specified for high
performance applications.
In addition to fan speed,
the operating temperature of the motor is determined by how
effectively the heat generated in the motor can be conducted
to surfaces which are in contact with the cooling medium (generally
air) and the ability to transfer this heat via convection
to the cooling medium. In a conventional totally enclosed
fan cooled motor the heat must be transferred from the laminated
steel stator core to the cast iron frame and finally to the
air. Since the fan is located opposite the drive end of the
motor, there is generally greater air flow and heat transfer
at one end of the motor than the other. Square laminated frame
AC motors have been offered by a variety of manufacturers
as a method to improve heat transfer. The laminated frame
design eliminates the stator-to-frame interface and provides
a more direct and effective heat transfer path to the cooling
air while integral cooling ducts trap the air in contact with
the frame along the motor's length. This laminated frame construction
has been common in variable speed DC motors for over twenty
years.
An offshoot of motor cooling
is the need to protect the motor should the motor cooling
system fail. While thermostats and thermistors are not common
in fixed frequency AC motors they should be required for variable
speed applications. A standard AC motor operates at a fixed
speed on a well-defined power supply which allows the shaft
driven fan to provide adequate cooling air in all normal circumstances.
By design a variable frequency control will allow the motor
to operate at very low speeds where little or no cooling is
provided. This might occur during maintenance, jog, or threading
operation for example. On the other hand, if a separately
powered blower is provided the drive motor must be protected
from a potential blower failure. As is the case with DC motors,
over temperature protection is recommended.
Figure 5 
Cooling Curves for TEFC Motors
--------------------------------------------------------------------------------
Disadvantages of Oversizing
(Derating) Motors
In applying variable frequency
controllers attempts are often made to use either "inplace"
AC motors, or standard sinewave power designs. To do this,
and operate across a speed range the motor is often oversized
relative to the rating required by the application. This can
sometimes be done successfully, but there are a number of
potential pitfalls. These can range from something as basic
as a motor insulation system which is fine on sinewave power,
but inadequate for the voltage and current waveshapes on the
controller, to drive system instability due to a lack of damping.
The oversized motor will have correspondingly higher rotor
inertia, which could lengthen acceleration and deceleration
times and reduce process productivity. Also, since no load
current tends to be a fairly constant percentage of full load
current within a motor product line, the higher no load current
of a derated motor could result in lower power factor and
higher current at the load point required by the application.
This current may exceed the capability of the variable frequency
controller requiring a costly oversizing of the controller
as well. A derated motor will have a lower nominal slip at
the application load than a matched motor, which can cause
problems either with load sharing in the case of multi-motor
drives, or with IET trips whenever the load changes quickly.
While it often appears to be economic to oversize a standard
motor to achieve a greater speed range, this course of action
should be approached cautiously while weighing all factors
of the desired performance of the drive.
--------------------------------------------------------------------------------
The Effect of Fast Power
Transistors
As power transistor technology
has evolved, there has been a proliferation of variable frequency
controllers operating at an AC input voltage of 460 V, using
these transistors as the power-switching device. As the transistor
manufacturers have continued to push toward devices with lower
losses and the capability of the higher switching rates, a
result has been very rapid transition times between the "off"
and "on" states. This is the case for both bipolar
(BJT) as well as insulated gate (IGBT) transistors.
The combination of fast transitions
(turn-on time) and the DC bus voltages of 460 VAC (input)
controllers results in the high "dV/dt" levels as
seen in Figure 6. What is typically referred to as dV/dt is
the time derivative of the voltage, or the slope of the voltage
versus time curve.
Figure 6 
Typical Transistors Transistion Voltage
Increasing the dV/dt levels
at the variable frequency controller output (and motor input)
can have effects which need to be considered in the design
of motors for such applications. The significance of these
effects can be shown by the following equation:
I = C x dV/dt
As can be seen from this
equation, as dV/dt increases, the capacitively coupled current
increases linearly with it. While items such as lead wires
and motors are not usually thought of in terms of capacitance,
three phase AC motor windings have a capacitance to ground
as well as between phases. The leads between the controller
and motor also exhibit similar effects. While these capacitance
values are normally considered negligible, given enough dV/dt,
it does not take much "C" to get quite a bit of
"I".
A second way of viewing the
high dV/dt levels is to use transmission line theory to compute
the voltage distribution due to the propagation of the steep
wavefront. This involves careful modeling of the leads and
motor windings as well as transition points such as conduit
box connections. Reflected as well as incident wavefronts
must be computed and combined. This type of analysis will
not be described in this paper. Analyses done by this methodology
are susceptible to errors due to many things including the
choice of appropriate complex impedance models for circuit
components. Generally, the results of this type of analysis
have indicated that the first length of wire in a motor will
see higher voltages than will subsequent parts of the winding.
This type of modeling is typically used for the analysis of
high voltage surges incident on the terminals of very large
machinery.
Another result of the very
fast transition time of today's transistors is that the voltage
at the inverter output and the motor terminals is not the
same. The voltage waveshapes in Figures 7 and 8 demonstrate
typical differences. Using the transmission line model mentioned
above, the two major differences in these waveshapes can be
explained as follows. The impedance of the leads results in
the voltage wavefront being distributed to some extent across
those leads, softening the wavefront to a lower dV/dt level
at the motor terminals. Secondly, the termination of the transmission
line (leads) at the motor results in a reflected wave, producing
the overshoot and dampened oscillation seen in Figure 8. This
waveform could also be modeled as the response of an L, R,
C, circuit to an impulse input.
Figure 7 
Voltage Wavefront at Inverter Output
Figure 8 
Voltage Wavefront at Motor Terminals
The end result of these waveshapes
being applied to the motor terminals is increased stress on
the insulation system. Since these waveshapes do not exist
in sinewave applications it is clear that their effect has
not been considered in standard AC motor insulation systems.
The motor insulation system must be capable of withstanding
both the increased thermal stress as well as the capacitively
coupled currents and voltage stresses. Appropriate selection
of individual materials, properly integrated into a motor
insulation system is needed to withstand the demands of operation
on variable frequency controllers.
--------------------------------------------------------------------------------
Motor Flux Level
The fundamental frequency
component of the voltage output of a variable frequency controller
can be as high as the AC input to the controller. However,
this is often not achieved. In order to maintain PWM modulation
for example, the output voltage may be limited to 90-95% of
the incoming AC voltage. As long as this situation is recognized,
and appropriate design choices made, it does not usually present
a problem. When an existing motor design (expecting 460 V
at 60 Hz, for example) is applied to a controller which delivers
only 420V, there can be problems.
While NEMA standards for
fixed speed AC motors allow for a 10% voltage variation from
nominal, it is important to recognize that at 10% lower than
nominal flux, performance including the nominal HP rating
will vary. For example, it may require 10% more current than
nominal to deliver rated HP. While this additional current
is almost always available from the incoming line it may not
be available from the variable frequency controller. Users
that are familiar with static DC drives and their characteristics
in low line conditions may be unpleasantly surprised to find
that AC variable frequency controllers often do not provide
the same rating capability at low line conditions. Operation
of an AC motor at lower than nominal flux levels will result
in increased slip and rotor heating which is self compounding
and may lead to a thermal runaway condition. High efficiency
AC motors designed for sinewave operation are often particularly
susceptible to poor performance when the controller output
voltage is low, since they usually employ low flux density
designs at nominal terminal conditions.
--------------------------------------------------------------------------------
Measurements in a PWM Environment
Another effect of the rapid-rise-time
pulses which today's variable frequency controllers can apply
to motors is to challenge existing measurement tools and techniques.
The high dV/dt voltage pulses are themselves not trivial to
measure. Typically, an oscilloscope with a single shot bandwidth
greater than 10 MHz, plus a high voltage probe with high frequency
capability (carefully impedance matched) is required. Since
voltage isolators typically cannot faithfully reproduce these
waveshapes, the scope must be "floated" unless the
variable frequency controller is operating on a floating power
system. This then requires appropriate care to avoid electrical
shock to the operator.
Not only is measuring the
voltage pulses difficult, all other measurements on the equipment
are exposed to this high dV/dt environment. This requires
the use of equipment which has high noise immunity and excellent
rejection of common mode voltages. Common devices such as
thermocouple and tachometer readouts often "misbehave"
and provide unreliable readings if they are not capable of
faithful operation in these high dV/dt conditions. This effect
makes activities such as drive start-up and troubleshooting
difficult as specialized equipment is required to take even
basic measurements.
Controlled
Pumpig.pdf
