Beruflich Dokumente
Kultur Dokumente
The dc output voltage is always less than the dc input voltage. The
converter is known also as a chopper.
i=
As L/RL >> T , the current ripple is small, so the load voltage, iRL , is essentially
constant at the mean voltage level, VDC . Hence,
The peak-to-peak load voltage ripple, V, is the current ripple multiplied by the
load resistance.
A high switching frequency causes higher losses, giving a trade off with the
inductor and capacitor size.
To obtain the condition for the onset of discontinuities conduction, consider the
inductor current at the boundary of continuous conduction.
During the rise and fall in current the di/dt 's are linear as the driving voltage
across the inductor is constant provided that the capacitor maintains V O
essentially constant.
During current build up the voltage across the inductor is VDC - VO , and during
discharge it is - VO . (Neglect the diode voltage for this analysis).
Build up ( Transistor on)
The highest critical average current, IOBmax , occurs at duty cycle, = 0.5.
When the MOSFET is turned on, the current builds up in the inductor. The
MOSFET is turned off, the inductor discharges into the load capacitor through
the diode. The time constant RLC is long compared to the switchiong period T
and all components are ideal.
9.3.2 Analysis for continuous conduction mode
During discharge, the inductor voltage is (VDC - VO) . Again in integral form
Equating
Hence
This is plotted in the form for a switched mode power supply (i.e. Vo is kept
constant by the controller).
The output voltage magnitude may be greater or smaller in than that of VDC.
The circuit is often used to obtain a negative supply rail. Again a controller
is essential.
10 Bridge converters
10.1 The full bridge dc-dc converter
The full bridge, or H bridge, converter permits reversible power flow, unlike the
converters described in lecture 8. The full bridge converter has several areas of
applications, namely dc motor drives, dc to ac inversion with a medium
frequency sine wave output and high frequency dc to ac conversion as part of a
switch mode power supply.
The circuit supplying a simple L/R load
There are two legs and each leg is made up of two switches and two diodes in
inverse parallel. Here n-channel MOSFETs are shown and the diodes are the
body diodes of the MOSFETs, which are good enough if the switching speed is
modest.
A number of ways of controlling the bridge exist. It is usual to have one of
the two transistors in each leg conducting at a given time, although in
practice a dead time when both are off is allowed to avoid shorting the dc
rail. The effect of too short a dead time is a shoot-through current, which
leads to high losses, if not destroying the devices. It is possible for both
transistors to be turned off in a leg as the freewheel diodes will pass any
inductive load current, but this mode of operation is unattractive as the
output voltage of the leg would depend on the direction of the load current,
causing difficulties in the control of the bridge.
The average inductor current at the boundary, IOB =
T1 , T2 on
T1 , T2 off
Providing one transistor in a given leg is always on, and the current is
continuous, the mean output voltage is given switching pattern of the upper
and lower switches. This is the most practical and common way of
controlling such as bridge and is identical to the way logic gates work.
The average value of vA ,
VA =
where
So the full bridge leg outputs can be considered separately, as step down
converters and in a bridge they allow for full four quadant operation:
Assuming that the dead time is negligible, the two duty cycles are related:
B =
)
So
10.2.2 Generation of
The control of the pulse width modulation often uses a triangular waveform of
amplitude V t :
The duty cycle A is set by the time for which the triangular waveform does not
exceed a control voltage vc. The on time, t is
is
T
V t4
The result shows that the output voltage is a linear function of the control
voltage.
Clearly the load voltage jumps between +VDC and -VDC , hence the name bipolar
voltage switching. The output current can be positive or negative with either
polarity of output voltage, so the power flow is reversible. As in the chopper,
inductance in the load circuit is attractive to smooth the current.
Various control methods are possible. Pulse width or duty ratio modulation
is preferred generally as the harmonic content is fairly stable and easy to
filter. A pwm frequency of greater than 20kHz has the advantage of being
out of the audible range. Many microcontrollers contain a pwm generation
block.
Applications:
When T1 is on, D1 passes the load current and vA is VDC; when T4 is on vA is 0V,
so the values of vA, and vB, are same whatever the direction of I L.
As the two legs are operating independently, a simple example is T 1 held in the
on-state and T2 used to modulate the voltage across the load.
More commonly the two legs do similar things with a phase displacement:
For example, the pulse width modulation strategy may use a triangle wave with
two control voltages +vc and -vc. The switches in leg A are controlled by vc and
those in leg B are controlled by -vc.
T1 is on if
As before,
T3 is on if
Period I:
T1 and T2 on
Period III: T1 and T2 on
The switching frequency of each leg is the same and the same as for bipolar
switching as it is given by the frequency of the triangle waveform.
Here the benefit of the triangle carrier can be appreciated. A clear phase
displacement in the leg switching is seen. The sawtooth carrier common in
microciontrollers would not give the effective doubling of the switching
frequency seen here by the load. Here, the ripple at the load is much lower.
Consequently, the boundary between continuous and discontinuous
conduction will be at a lower current.
The average inductor current at the boundary, IOB =
With bipolar switching, the amplitude of the ac output can only be controlled by
varying the dc input voltage, which could, for example, be derived from a
step-down converter. The frequency can be varied by changing the frequency of
the controlling triangular wave. The output contains all odd harmonics.
With Unipoar switching, the ac output and its harmonic current can be varied by
keeping the duty ratio B the same as A but the phase between the leg
switching is shifted away from 180O (or 0O).
The voltage across the load is half the value obtained from the full bridge. The
mid-point of the capacitors is at VDC/2.
10.5 Summary
The Bridge converter circuits offer a great amount of flexibility. Their use in
dc-dc conversion is widespread at powers above one or two hundred watts. They
are widely used in dc servo motor drives. The use of full and half bridges to
generate ac is also widespread, particularly for ac motor drives, isolated switch
mode power supplies and RF heating.
Operation in the 'square-wave' mode is simple, but the harmonic content of
the output is high. For some applications at high frequencies, > 20 kHz,
such as switch mode power supplies and RF heating, the square waveform is
acceptable or the load can be tuned to minimise harmonic currents (see later
Lectures). For ac motor drives, operating at around 50 Hz, filtering
components are very bulky and expensive and harmonics cause severe
losses. In these systems, pulse width modulation (pwm) schemes with
sinusoidal references are adopted to produce an output which is closely
sinusoidal.
The average inductor current at the boundary, IOB =
W A
2 V I
kfJBpk
W is the winding window area, A is the flux path area and k is the packing
factor for the winding. W amd A are orthogonal, so indicate the volume, where
their ratio is fixed by the core shapes available. Clearly a high frequency
reduces the volume and weight of the transformer.
With some circuits for example the flyback converter, the mean flux in the
transformer is not zero. This must be allowed for on the design, typically by
putting an air gap in the magnetic path. This results in poor utilisation of the
core. Some cores are manufactured with an airgap and design curves are
used to determine the winding details.
The iron core loss is represented by a resistor, Rm. Using the correct ferrite for
the frequency of operation allows Rm to be neglected.
For low output voltages a split secondary as shown is preferred as only one
diode is in series with the load at any given time. An inductor must be used
as the first smoothing element (or design in transformer leakage inductance).
Follow with capacitor smoothing.
For high voltage outputs a bridge rectifier gives the smallest reverse voltage
across an individual diode and better use of the transformer winding. Again,
add capacitor smoothing.
11.2.2 The full bridge circuit
The full bridge is capable of supplying the transformer primary with twice
the voltage when compared to the half bridge, and the output voltage can be
controlled by unipolar switching, but needs four transistors. Care must be
taken to ensure that there is no dc current in the transformer. A dc current
would be wasted power and could also saturate the transformer.
Note the dot convention, which indictes the voltage sense of the windings.
When the transistor is on, the secondary voltage is negative with respect to the
diode D2. When the transistor turns off, a current into the dot is required to
maintain the transformer flux, so diode D 2 becomes forward biased and a
secondary current flows.
The use of a transformer allows either side of the output to be earthed,
enabling a supply of either polarity to be obtained as well as allowing the
switch to be at ground. It may also be used to reduce the voltage stress on
the transistor.
11.3.2Analysis
Clearly, the energy stored in the inductor is maintained at the switching instant.
For analysis, the transformer is better represented by the magnetising inductance
and an ideal transformer (with the magnetising branch on the primary side).
current decay :
t T
The current in LO varies from IO1 as the transistor turns on to IO2 where the
transistor turns off. The current charge, IO = IO2 - IO1
This is similar to the result obtained for the buck-boost converter, using an
analysis of the current, but the output voltage in the flyback converter is also
transformed by N2/N1 allowing better choices in the design.
The voltage across the switch when it is off is
The voltage across the switch here is lower than that for a given output
voltage with the buck-boost by virtue of the transformer action. This
overcomes the main drawback of the buck boost, making it the better circuit
for high ratios of step up conversion. Like the buck boost it is usually used
in the discontinuous current mode, under the control of a dedicated control
IC, so the simple equation does not then apply.
The average inductor current at the boundary, IOB =
Transistor on
t T:
The primary voltage is VDC. D1 conducts and supplies load current. The
transistor current will consist of two components, i1 where i1 = N1/N2 i2 , and iO
which is the magnetising current. Appropriate transformer design ensures that
i O < 10% i 1 .
Transistor off,
t T
The transformer flux decays, reversing the transformer voltage so that diode D3
becomes forward biased. A current flows in the tertiary winding. In this period
the primary voltage is determined by the tertiary winding and the turns ratio,
(-N1/N3)VDC . This period is used to demagnetise the transformer.
For no net flux in the core, the increase in current when the transistor is on must
equal the decrease in the off period.
Hence
max
is then
number of turns on the tertiary winding (if it exists) equals the number of
turns on the primary.
A bridge with only two switches and two freewheel diodes. The switches are
operated simultaneously (bipolar switching). The circuit eliminates
demagnetising windings, since D 3 and D 4 take over the magnetising current
when T 1 and T2 are switched off, so the reverse applied voltage demagnetises
the core. Max duty ratio of 50%.
This is a rather tedious circuit as a high side gate drive is required, but
otherwise simple and efficient with excellent Mosfets available below 60V.
The duty cycle of the switches is fixed at 0.5. Obviously the two transistors
must not conduct simultaneously. On turning off T1 , D2 takes the magnetising
current for a short period until the transformer current reverses.
It is difficult to avoid the transformer saturating, as an exactly 50% duty
cycle is hard to achieve.
The basic output voltage is related to the input voltage by
N2
VO N1 VDC
12 Switch-mode dc to ac inverters
Many applications, notably ac motor drives, require a variabe voltage variable
frequency ac source which has a lower harmonic content than can be obtained
by the simple square wave operation of a bridge converter as described in
lecture 10. An effective solution is to use sinusoidal pulse width modulation.
Waveform generation:
1. the higher harmonics in the output are sidebands centred around the
switching frequency its multiples
During this time, diode D1 will be turned off and the load current continues
through Df . The voltage on the transistor is (1+N1/N3)VDC.
Known as over modulation or pulse dropping, the limit is square wave operation
as in section 9.3.
The output at the fundamental frequency no longer varies linearly with the
magnitude of vc .
12.1.4 Implementation
The features described above are most conveniently combined using
microcontrollers or FPGAs. The odd integer synchronised ratio of fS to f1 can
then be produced easily.
Various alternative methods for producing sinusoidal pwm have been
implemented, the most common is based on pre-calculated waveforms
which explicitly eliminate particular harmonics.
The average inductor current at the boundary, IOB =
where n and k are again integers. Both approaches give the same amplitude of
the fundamental for a given modulation ratio m.
The effect is better than it appears in the waveforms: All the components at
and around the modulating frequency are eliminated.
An alternative unipolar scheme employs reference waveforms staggered by
120O.
The effective doubling of the switching frequency is retained, but the output
voltage magnitude (voltage gain) is reduced to 0.866 at m = 1 . The harmonic
content of the output voltage can be reduced by synchronising fs and f1 , making
mf odd integer, as before and a multiple of 3.
All Third harmonics in the phase voltages cancel in the line voltage due to the
120O difference VA to VB
The unipolar waveforms above, with 120O between the reference waveforms,
provide the three modulated outputs, A, B, C.
Thus three line voltages can be constructed
23
Th
VDC 5 7 1 1 11 1 13 ....