You are on page 1of 3

Welcome to part 6 of this series of videos.

Now, we're going to expand on the


introductory descriptions in part 5 and take a detailed look at how the phase shift
full bridge works. I showed this animation at the end of part 5, but I think it's
worth repeating here. The graphic shows the gate drives to QA and QB in red and
blue on the top and those to QC and QD in red and blue in the middle.

The resulting voltage across the transformer is shown in brown at the bottom. The
switching waveforms are all at 50% duty cycle, less 1% or 2% to allow time for the
ZVS transitions. OUTA and OUTB are the fixed reference pair and the control shifts
the phase of OUTC and OUTD with respect to the reference pair, in order to control
the duty cycle seen by the transformer.

Energy transfer from primary to secondary happens when the diagonal pairs of
switches are on, QA and QD or QB and QC. The transformer duty cycle is simply the
proportion of the switching cycle for which voltage is applied to the transformer.
Now, in the next eight slides, I will describe a typical switching cycle of the
phase shift full bridge. We'll start at the beginning of a switching cycle, where
the diagonal pair QA and QD are on and energy is being transferred from the input
to output.

This is the first of the two active states in the switching cycle. The red and blue
arrows indicate current flow in the primary and secondary circuits. Now, QA and QD
are on for time T-ON, which is determined by the controller. And there is a
positive voltage across the transformer primary, and the voltage at the secondary
is rectified by the SR-QF. The current in the output inductor is increasing as the
energy is transferred from primary to secondary.

At some time, the PWM controller turns QD off, and this terminates the energy
transfer or active interval. This starts a positive going ZVS transition on the
right leg and the system changes from actively transferring energy from the primary
to the secondary to a passive or freewheeling state at the end of the ZVS
transition. The ZVS transition on the right leg is driven by the energy in the
leakage, magnetizing, and output inductances, which combine to charge the
capacitance at Node B. And this node swings positive from 0 volts to V-IN when QD
is turned off, where it is clamped to about V-IN plus 0.7 volts when the body diode
of QC turns on.

There is a delay time between QD turning off and QC turning on, shown here as
DELCD. And this allows time for the ZVS transition to happen. The MOSFET QC is
turned on at the end of the DELCD delay, and there are no switching losses, because
there is 0 volts across it that time. We shall see later that the energy available
to drive the transition at Node B on the right, or AP leg, is greater than not
driving the transition at Node A on the left, or PA leg.

This means that the transitions on the right leg are faster than those at Node A on
the left leg. The primary and secondary currents at the end of the ZVS transition
interval are shown in red and blue. At the end of the ZVS transition, QE is turned
on, and both the primary and secondary of the transformer are short circuited. At
this time, there is no voltage across the transformer primary or secondary, and we
know that V is LD-IDT. So if V equals to 0, then DIDT must also equal to 0.

This means that the circulating current to the primary is preserved and is
available to drive a ZVS transmission at the end of this interval. This ability to
achieve ZVS is a key feature of the phase shift full bridge that differentiates it
from the PWM full bridge. Losses in circuit resistances mean that the circulating
current does decay a little during this interval, but this is not usually
significant.

There is no energy transfer from input to output during this time, so that the
bridge is in a passive state. This interval is also sometimes called a freewheeling
interval. In the preceding energy transfer interval, the output current was carried
by QF. And now during this subsequent passive interval, the average current remains
in the QF and only the change in the inductor current splits between the two
windings.

This is a consequence of the fact that there is no voltage across the winding
inductances to force a current out of one winding and into the other. This
asymmetry means that RMS losses in the transformer secondary are higher than they
would be if the current is shared equally. This interval ends when the PWM clock
starts the second energy transfer cycle.

The PWM clock turns QA off to initiate a negative going ZVS transition on the left
leg. As with the earlier ZVS transition, there is a delay time, DELAB, between QA
turning off and QB turning on. This delay allows time for the energy stored through
the leakage inductance to charge the capacitance at Node A negatively, until it is
tamped by the body diode of QB at about minus 0.7 volts.

The MOSFET QB is then turned on at the end of the DELAB interval, with 0 volts
across it and no switching loss. Remember that the ZVS transition on the right leg
is driven by the energy of the leakage, magnetizing, and output inductances.
However, only the energy in the leakage inductance is available during the left leg
ZVS transmission, because the transformer secondary is short circuited.

This means that the left leg transitions take longer than the right leg
transitions, and the UCC28951-Q1 allows the designer to set different DELAB and
DELCD times for this reason. The fact that there is less energy available also
means that at light loads, it is more difficult to maintain ZVS on the left leg
than it is on the right leg.

QF is turned off during this interval and this releases the short circuit on the
transformer secondary, in readiness for the second energy transfer interval. So
this is the second energy transfer interval during the switching cycle. In this
slide, QB and QC are on for a time T-ON, which is determined by the control loop.

The transformer polarity is now opposite to that in the first energy transfer
interval, and there is now a negative voltage across the transformer primary. The
voltage at the second interval of the transformer is rectified by the SRQE, and the
current in the output inductor is increasing as the energy is transferred from
primary to secondary. You can also see that the switching frequency seen by the
output inductor is twice that seen by the transformer.

This can cause confusion, unless you are very clear about what you mean when you
describe the switching frequency of a phase shift full bridge. The second energy
transfer interval is terminated when QC is turned off by the PWM controller. This
initiates a negative going ZVS transition on the right leg, and the system changes
from actively transferring energy from the primary to the secondary to a passive
state, where there is no energy transfer.

As before, the DELCD delay allows time for the ZVS transition. This transition is
driven by the energy of the leakage, magnetizing, and output inductances, which
combined to charge the capacitance at Node B. As a result, this node swings
negative from VN, and is counted about minus 0.7 volts when the body diode of QD
turns on. Then, at the end of the DELCD delay, the MOSFET QD is turned on, and this
is done with 0 volts across the MOSFET and no switching losses.

At the end of the ZVS transition, QF is turned on, so that both the primary and
secondary of the transformer are short circuited. As before, when QA and QC were
both on, the circulating current of the primary is preserved and is available to
drive a ZVS transition at the end of this interval. As in the earlier freewheeling
interval, current is supplied by LOUT through the transformer secondary, although
as before, the current flows asymmetrically in the transformer secondaries.

This interval ends when the PWM clock starts the second energy transfer cycle. The
PWM clock starts the second energy transfer interval by turning QB off, and this
starts a positive-going ZVS transition on the left leg. As before, the delay time,
DELAB, allows time for the energy stored in the leakage inductance to charge the
capacitance at Node A. And this node swings positive from 0 volts when QB is turned
off.

The transition is clamped at about VN plus 0.7 volts when the body diode of QA
turns on. QA is then turned on at the end of the DELAB interval, but it does so
with 0 volts across it and no switching loss. QE is turned off during this
interval, and this releases the short circuit on the transformer secondary. And
this completes the switching cycle, and the process repeats indefinitely.

Now, please feel free to pause this slide, to study it if you wish. But the main
characteristics to remember are that all of the SR transitions are 0 voltage
switched and that they take longer as the load current increases. This behavior is
different to that of the primary side switched nodes, where the transitions take
less time as current increases.

The main thing to note on the active passive SR transition is that one of the SRs
is already on. The transformer secondary voltage falls to 0 and then the second SR
is turned on. This happens at 0 volts and the transformer secondary is then short
circuited by the two SRs. The time needed to achieve ZVS on both the
passive/active, and active/passive legs depends on the transformer current. Some
controllers change the delay times as a function of the current. This feature,
which we call adaptive delays, is provided on the UCC28951-Q1 and UCC2895-Q1
controllers.

Note that the delays on the primary side switches decrease as the transformer
current increases, and that the delays on the secondary switched increase as the
transformer current increases. SR disable is the ability to disable the SRs and
revert to diode rectification at light loads. This prevents reverse currents in the
resonant tank and improves light load efficiency, because it eliminates the losses
due to switching the MOSFETs. As regards bidirectional operation, well, the phase
shift full bridge isn't well-suited to bidirectional option, but we do have some
examples, some of which are listed here.

The next video of this training series is Part 7, which is where we look at how a
high power on board charger might be designed using the ideas discussed previously.

You might also like