U.S. patent application number 16/387142 was filed with the patent office on 2020-10-22 for control method and system for prevention of current inversion in rectifiers of switching converters.
The applicant listed for this patent is STMicroelectronics S.r.l.. Invention is credited to Alberto Iorio, Jeanpierre Vicquery, Emilio Volpi.
Application Number | 20200336071 16/387142 |
Document ID | / |
Family ID | 1000004054515 |
Filed Date | 2020-10-22 |
![](/patent/app/20200336071/US20200336071A1-20201022-C00001.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00000.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00001.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00002.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00003.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00004.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00005.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00006.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00007.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00008.png)
![](/patent/app/20200336071/US20200336071A1-20201022-D00009.png)
View All Diagrams
United States Patent
Application |
20200336071 |
Kind Code |
A1 |
Iorio; Alberto ; et
al. |
October 22, 2020 |
CONTROL METHOD AND SYSTEM FOR PREVENTION OF CURRENT INVERSION IN
RECTIFIERS OF SWITCHING CONVERTERS
Abstract
A method of controlling synchronous rectification transistors in
a switching converter includes sensing a drain-to-source voltage
across each synchronous rectification transistor each switching
half-cycle of the switching converter. An average of the sensed
drain-to-source voltage is calculated for each synchronous
rectification transistor over N prior switching half-cycles. A load
current transient in the switching converter is sensed based on the
sensed drain-to-source voltage of each synchronous rectification
transistor and the calculated average of the sensed drain-to-source
voltage for each synchronous rectification transistor over the N
prior switching half-cycles.
Inventors: |
Iorio; Alberto; (Aosta,
IT) ; Volpi; Emilio; (Charvensod, IT) ;
Vicquery; Jeanpierre; (Brusson, IT) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
STMicroelectronics S.r.l. |
Agrate Brianza |
|
IT |
|
|
Family ID: |
1000004054515 |
Appl. No.: |
16/387142 |
Filed: |
April 17, 2019 |
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H02M 1/088 20130101;
H02M 3/33592 20130101 |
International
Class: |
H02M 3/335 20060101
H02M003/335; H02M 1/088 20060101 H02M001/088 |
Claims
1. A method of controlling synchronous rectification transistors in
a switching converter, the method comprising: sensing a
drain-to-source voltage across each synchronous rectification
transistor each switching half-cycle of the switching converter;
and calculating an average of the sensed drain-to-source voltage
for each synchronous rectification transistor over N prior
switching half-cycles, wherein N is an integer greater than 0.
2. The method of claim 1, comprising: detecting a load current
transient in the switching converter based on the sensed
drain-to-source voltage of each synchronous rectification
transistor and the calculated average of the sensed drain-to-source
voltage for each synchronous rectification transistor over the N
prior switching half-cycles, the detecting a load current transient
including: calculating a peak voltage reference threshold for each
synchronous rectification transistor having a value based on the
calculated average of the sensed drain-to-source voltage for the
synchronous rectification transistor; and comparing the sensed
drain-to-source voltage of each synchronous rectification
transistor to the corresponding peak voltage reference threshold
for the synchronous rectification transistor.
3. The method of claim 2, wherein calculating the peak voltage
reference threshold for each synchronous rectification transistor
comprises: calculating a peak sensed drain-to-source voltage from
the calculated average of the sensed drain-to-source voltage for
the synchronous rectification transistor; and multiplying the peak
sensed drain-to-source voltage by a threshold adjustment factor to
generate the peak voltage reference threshold.
4. The method of claim 2, wherein detecting the load current
transient further comprises, for each synchronous rectification
transistor: comparing the sensed drain-to-source voltage to the
peak voltage reference threshold during a conduction time of the
synchronous rectification transistor; determining whether the
sensed drain-to-source voltage reaches the peak voltage reference
threshold within a certain percentage of the conduction time; and
detecting the load current transient in response to the sensed
drain-to-source voltage not reaching the peak voltage reference
threshold within the certain percentage of the conduction time.
5. The method of claim 4, further comprising multiplying the
conduction time by a conduction time adjustment factor to determine
the certain percentage of the conduction time.
6. The method of claim 5, wherein calculating of the average of the
sensed drain-to-source voltage for each synchronous rectification
transistor over N prior switching half-cycles comprises calculating
the average of the sensed drain-to-source voltage during the
conduction time of the synchronous rectification transistor.
7. The method of claim 1, further comprising controlling the
switching converter through an adaptive turn-off control
algorithm.
8. The method of claim 7, wherein the adaptive turn-off control
algorithm generates a programmable delay that is utilized in
controlling each synchronous rectification transistor, and wherein
the method further comprises resetting the programmable delay to
zero in response to detecting a load current transient in the
switching converter.
9. The method of claim 1, wherein the switching half-cycle of each
synchronous rectification transistor begins in response to the
sensed drain-to-source voltage of the synchronous rectification
transistor reaching a first threshold value and ends in response to
the drain-to-source voltage reaching a second threshold value.
10. The method of claim 9, wherein the first threshold value is
less than the second threshold value.
11. A control circuit configured to be coupled to synchronous
rectification circuitry of a switching converter, the control
circuit configured to generate control signals to control switching
of at least one synchronous rectifier transistor in the synchronous
rectification circuitry, each at least one synchronous rectifier
transistor including a control node and signal nodes, and the
control circuit configured to sense a voltage across the signals
nodes of each at least one synchronous rectifier transistor for a
portion of each switching cycle of the switching converter in which
the at least one synchronous rectifier transistor is activated, and
the control circuit configured to determine for each at least one
synchronous rectifier transistor, over N prior portions of
switching cycles in which the at least one synchronous rectifier
transistor is activated, N being an integer greater than 0, an
average of the sensed voltage across the signal nodes of the at
least one synchronous rectifier transistor.
12. The control circuit of claim 11, wherein the control circuit is
configured to generate first and second control signals to control
switching of first and second synchronous rectifier transistors,
respectively, and wherein the portion of the switching cycle of the
switching converter includes a first switching half-cycle in which
the first synchronous rectifier transistor is activated and a
second switching half-cycle in which the second synchronous
rectifier transistor is activated.
13. The control circuit of claim 12, wherein the control circuit
comprises: an average calculation circuit coupled to at least one
of the signal nodes of each of the first and second synchronous
rectifier transistors, the average calculation circuit configured
to sense the voltage across the signals nodes of the first and
second synchronous rectifier transistors during a conduction time
of the first and second synchronous rectifier transistors, and
configured to calculate a corresponding first and second average
voltage of the sensed voltage over the N prior first and second
switching half-cycles for the first and second synchronous
rectifier transistors, respectively; and a peak calculation circuit
coupled to the average calculation circuit to receive the
calculated first and second average voltages, and the peak
calculation circuit configured to generate first and second peak
voltage reference thresholds based on the calculated first and
second average voltages, respectively.
14. The control circuit of claim 13, wherein the peak calculation
circuit is coupled to multiply the first and second calculated
average voltages by a threshold adjustment factor to generate the
first and second peak reference voltage thresholds.
15. The control circuit of claim 14, wherein the control circuit
further comprises first and second comparators, the first and
second comparators each including a first input coupled to one of
the signal nodes of the first and second synchronous rectifier
transistors, respectively, and the first and second comparators
each including a second input coupled to receive the peak voltage
reference threshold generated for the first synchronous rectifier
transistor and the second synchronous rectifier transistor,
respectively, and wherein the first and second comparators are
configured to generate first and second comparison output signals,
respectively, each of the first and second comparison output
signals indicating whether a load transient has occurred in a
current first switching half-cycle and second switching half-cycle,
respectively.
16. A switching converter, comprising: switching circuitry
including at least one switching element; synchronous rectification
circuitry coupled to the switching circuitry, the synchronous
rectification circuitry including at least one synchronous
rectifier transistor, each at least one synchronous rectifier
transistor including a control node and signal nodes; and a control
circuit coupled to each at least one synchronous rectifier
transistor, the control circuit configured to: sense a voltage
across the signals nodes of each at least one synchronous rectifier
transistor during each portion of switching cycle of the switching
converter in which the at least one synchronous rectifier
transistor is activated; and calculate an average of the sensed
voltage for each at least one synchronous rectifier transistor over
N prior portions of the switching cycle in which the at least one
synchronous rectifier transistor is activated, wherein N is an
integer greater than 0.
17. The control circuit of claim 16, wherein the control circuit is
configured to detect a load current transient in the switching
converter based on a comparison of the sensed voltage of each at
least one synchronous rectifier transistor during the portion of
the switching cycle in which the at least one synchronous rectifier
is activated and the calculated average of the sensed voltage for
the at least one synchronous rectifier transistor over the N prior
portions of the switching cycle in which the at least one
synchronous rectifier transistor is activated.
18. The control circuit of claim 16, wherein each at least one
synchronous rectifier transistor comprises a MOSFET, and the
synchronous rectification circuitry includes first and second
MOSFETs.
19. The control circuit of claim 16, wherein the control circuit is
further configured to implement an adaptive turn-off control
algorithm to control the switching of the first and second
MOSFETs.
20. The control circuit of claim 16, wherein the switching
circuitry comprises LLC resonant converter circuitry.
Description
BACKGROUND
Technical Field
[0001] The present disclosure relates generally to a control method
for a rectifier of a switching converter, and more specifically to
the prevention or reduction of current inversion in the synchronous
rectifier of a synchronous switching converter.
Description of the Related Art
[0002] Resonant converters are a large class of forced switching
converters characterized by the presence of a half-bridge or a
full-bridge circuit topology. In the half-bridge version, for
example, the switching elements comprise a high-side transistor and
a low-side transistor connected in series between an input voltage
and ground. A square wave having a high value corresponding to the
power supply voltage and a low value corresponding to ground may be
generated by conveniently switching the two transistors.
[0003] The square wave generated by the half-bridge is applied to
the primary winding of a transformer by means of a resonant circuit
which comprises at least one capacitor and one inductor. The
secondary winding of the transformer is connected with a rectifier
circuit and a filter to provide an output direct voltage depending
on the frequency of the square wave.
[0004] At present, one of the resonant converters most widely used
is the LLC resonant converter. This name derives from the fact that
the resonant circuit employs two inductors (L) and a capacitor (C).
A schematic circuit of an LLC resonant converter is shown in FIG. 1
and comprises a half-bridge of MOSFET transistors M1 and M2, with
respective body diodes Db1 and Db2, coupled between an input
voltage Vin and ground GND and driven by a driver circuit 3. The
common terminal between transistors M1 and M2 is connected to a
resonant network 2 comprising a series of a first inductance Lr, a
second inductance Lm and a capacitor Cr; the inductance Lm is
connected in parallel to a transformer 10 comprising a secondary
winding connected to a parallel connected capacitor Co and
resistance Rout by means of the rectifier diodes D1 and D2. The
output voltage Vo of the resonant converter is the voltage across
the parallel connected capacitor Co and resistance Rout, while the
output or load current Io/Iload flows through the resistance
Rout.
[0005] These resonant converters are characterized by a high
conversion efficiency (>95% is easily achievable), an ability to
work at high frequencies, low generation of EMI (electro-magnetic
interference).
[0006] In current types of converter circuits, a high conversion
efficiency and high power density are desired, as in the case, for
example, of the AC-DC adaptors of notebooks. LLC resonant
converters are at present the converters that best meet such
desires. The maximum efficiency achievable is limited, however, by
losses in the rectifiers on the secondary side of the converter,
which account for over 60% of total losses.
[0007] To significantly reduce the losses connected to secondary
rectification, a "synchronous rectification" technique is used in
which rectifier diodes (e.g., D1 and D2 of FIG. 1) are replaced by
power MOSFETs. The power MOSFETs have a suitably low on-resistance
R.sub.dson such that a voltage drop across each MOSFET is
significantly lower than the voltage across each rectifier diode
D1, D2. These power MOSFETs are driven in such a manner to be
functionally equivalent to the rectifier diodes D1, D2. This
synchronous rectification technique is widely adopted in
traditional converters, especially in flyback and forward
converters, for which there also exist commercially available
dedicated integrated control circuits. This synchronous
rectification technique has also been utilized in resonant
converters as well, in particular in LLC converters, in order to
improve the efficiency of such converters.
[0008] FIG. 2 shows the converter of FIG. 1 except with the
rectifier diodes D1, D2 having been replaced with synchronous
rectification circuitry. In this case, in the place of the
rectifier diodes D1 and D2 there are two synchronous rectifier
transistors SR1 and SR2, suitably driven by two signals G1 and G2
generated by a synchronous rectifier driver 80. The transistors
SR1, SR2 are referred to as either synchronous rectifier
transistors or synchronous rectification transistors in the present
description. The transistors SR1, SR2 are connected between the
terminals of the center tap of the center-tapped secondary winding
and ground GND, while the parallel connected capacitor Co and
resistance Rout are disposed between a rectifier output node OUTN
coupled to the other terminals of the center-tapped secondary
windings and ground GND. From a high-level functional viewpoint
there is no difference between the synchronous rectification
circuitry of FIG. 2 compared to the rectification circuitry formed
by the rectifier diodes D1, D2 in FIG. 1.
[0009] The transistors SR1 and SR2 have respective body diodes Dbr1
and Dbr2, and are both driven by the synchronous rectifier driver
80. The output voltage Vo of the resonant converter is the voltage
across the parallel connected capacitor Co and resistance Rout,
while the output current Io/Iload flows through the resistance
Rout. In the present description where more than one of a given
component exists, the components are typically referred to using a
reference letter followed by a reference number, such transistors
SR1, SR2 and body diodes Dbr1, Dbr2. Moreover, when referring to a
specific one of these components in the present description, both
the reference letter and reference number will be utilized (e.g.,
SR1, SR2), while the reference letter alone will be utilized when
referring to all or any one of the components (e.g., SR and
Dbr).
[0010] In operation, the transistors SR1 and SR2 are driven in such
a manner to be alternatively turned-on at a certain frequency by
the synchronous rectifier driver 80. When the body diode Dbr1, Dbr2
of one of the transistors SR, SR2 starts conducting the relative
transistor is turned-on, while when the current is approaching to
zero the transistor is turned OFF. In this way the use of the
transistors SR1, SR2 results in a lower voltage drop than does the
use of the diodes D1, D2, reducing the power dissipation of the
rectification circuitry.
[0011] Particularly, as shown in FIG. 3, three phases of operation
are shown and labelled Phase A, Phase B and Phase C through the
letters enclosed in circles in the figure. The Phase A is activated
when a drain-to-source voltage Vdvs between the drain and source
terminals of one of the transistors SR1, SR2, for example
transistor SR1, is lower than a threshold voltage VTH1 and the
relative body diode Dbr1 starts conducting. When the voltage Vdvs
falls under or below a turn-on threshold voltage V.sub.TH_ON, after
a fixed delay time period T.sub.PD_ON, assuming the voltage Vdvs is
maintained under the turn-on threshold voltage V.sub.TH_ON, the
transistor SR1 is turned ON by the synchronous rectifier driver
80.
[0012] After the turn ON of the transistor SR1, operation in Phase
B begins and the voltage Vdvs has a value of Vdvs=-Rdson.times.Isr,
wherein R.sub.dson is the ON resistance of the transistor SR1, SR2
and Isr is the synchronous rectifier current flowing through the
electric path between the center-tap CT of the secondary winding of
the transformer and ground GND. This ON resistance may be
alternately referred to as R.sub.dson and R.sub.DS(on) in the
present description (i.e., R.sub.dson=R.sub.DS(on)).
[0013] When the voltage Vdvs has a value higher than a second
threshold voltage V.sub.TH-OFF, indicating the current Isr through
the transistor is approaching zero, operation in Phase C begins and
the transistor SR1 is turned OFF by the synchronous rectifier
driver 80. The respective body diode Dbr1 then conducts again and
the voltage Vdvs across transistor SR1 goes further negative as
illustrated in FIG. 3. The voltage Vdvs goes further negative in
response to transistor SR1 turning OFF because positive ISR1 is
still flowing but now flows through body diode Dbr1 and there is a
larger voltage drop across this body diode than the voltage drop
(Rdson.times.Isr) across the source and drain when the transistor
is turned ON. The voltage Vdvs then varies as shown in FIG. 3 and
begins increasing, and when the voltage Vdvs reaches the threshold
VTH2, Phase C ends. As soon as the voltage Vdvs reaches the
threshold voltage V.sub.TH_ON, the synchronous rectifier driver 80
turns ON the transistor SR2, and operation in Phases A, B and C as
just described for transistor SR1 occurs for transistor SR2.
[0014] The actual voltage Vdvs depends on parasitic elements of the
source and drain terminals of the transistor SR1, SR2 and of a
conductive trace or path on a printed circuit board (PCB) from the
drain terminal of the transistor SR, SR2 to the terminal of the
corresponding secondary winding of the transformer 10.
Particularly, the voltage Vdvs depends on the parasitic inductances
Lsource and Ldrain associated with the source and drain terminals
of the transistor SR1, SR2 and on the parasitic inductance Ltrace
of the conductive path of printed circuit board (PCB) from drain
terminal of the transistor SR1, SR2 to the corresponding terminal
of the secondary winding of the transformer 10. Therefore, due to
these parasitic inductances, the actual drain-to-source voltage
Vdvs ( t ) = - Rdson .times. Isr - ( Ldrain + Lsource + Ltrace )
.times. .differential. Isr .differential. t ##EQU00001##
for each of the transistors SR1, SR2. Thus, this parasitic or stray
inductance Lstray or L.sub.SR (L.sub.SR=Lsource+Ldrain+Ltrace)
makes the actual sensed voltage Vdvs1, Vdvs2 across each transistor
SR1, SR2 different from the ideal voltage drop value on across the
resistance R.sub.ds on presented by the transistor, where this
voltage is designated Vdvs-ideal in the following description.
[0015] The presence of this parasitic inductance L.sub.SR results
in an undesired earlier turn OFF of the transistors SR1, SR2 as
shown in FIG. 4 where the actual drain-source voltage Vdvs taking
into account the stray inductance L.sub.SR and the desired
drain-source voltage Vdvs-ideal are shown. This results in a
residual conduction time T.sub.diode of the body diode Dbr1 or Dbr2
increasing, causing a loss of efficiency indicated as a time LE in
FIG. 4 and which is due to the higher voltage drop across the body
diode Dbr1 or Dbr2 for the longer duration during the time LE. The
transistor SR1 turns OFF at a time t1 instead of turning OFF at a
time t2, where (t2-t1)=LE, resulting in this undesired increase in
the residual conduction time T.sub.diode of the body diode Dbr1.
For example, a typical actual residual conduction time T.sub.diode
of the body diode Dbr1, Dbr2 could be on the order of 1
microsecond, while a typical desired residual conduction time
T.sub.diode is approximately 60 nanoseconds.
[0016] There are known techniques to avoid the earlier turn OFF of
the transistors SR1, SR2 to improve the efficiency of converter.
One approach is to compensate the time delay due to the parasitic
inductance L.sub.SR by adding an RC filter between the synchronous
rectifier driver 80 and transistor SR1, SR2, and before the
transformer 10 as shown in FIG. 5. This RC filter approach of
compensating for the parasitic inductance L.sub.SR provides a
simple architecture with a consequent low cost in terms of silicon
area and good performance. Furthermore, the efficiency of this
solution is dependent on the residual conduction time T.sub.diode
of the body diodes Dbr1, Dbr2 which, in turn, depends on the ON
resistances of the transistors SR1, SR2, on the parasitic elements
of the transistors (Lsource, Ldrain) and the printed circuit board
(Ltrace) to which the transistors are connected, on temperature,
and on the slew rate of the current flowing through the
transistor.
[0017] Another approach that has been utilized to compensate for
the parasitic inductance LSR is referred to as an adaptive turn-off
control algorithm. In this approach, the idea is to take advantage
of the anticipatory effect or leading nature of the actual measured
or sensed drain-source voltage Vdvs caused by the parasitic
inductance LSR relative to the ideal drain-source voltage. This is
seen in FIG. 5 with the actual sensed
voltage Vdvs ( t ) = - Rdson .times. Isr - ( Ldrain + Lsource +
Ltrace ) .times. .differential. Isr .differential. t
##EQU00002##
where LSR=(Ldrain+Lsource+Ltrace) as defined above. The effect of
the parasitic inductance LSR is that the actual sensed voltage
Vdvs, as illustrated in FIG. 4, does not correspond to the ideal
drain-source voltage Vdvs-ideal of the transistor SR. This actual
sensed voltage Vdvs leads or anticipates the ideal drain-source
voltage Vdvs-ideal, meaning a zero crossing of the actual sensed
voltage occurs earlier in time than the actual zero crossing of the
ideal drain-source voltage. The adaptive turn-off control algorithm
utilizes the actual voltage Vdvs and detects a zero-crossing of
this signal, and measures the residual body diode Dbr conduction
time for the transistor SR. The algorithm then, cycle-by-cycle,
increases a programmable turn-off delay relative to the detected
zero-crossing to minimize the residual conduction time Tdiode of
the body diode Dbr and thereby reduce the resulting power losses in
the diode, improving the efficiency of the converter. The adaptive
turn-off control algorithm is described in more detail in U.S. Pat.
No. 9,356,535, which is incorporated herein in its entirety to the
extent not inconsistent with the specific teachings and definitions
set forth herein. Whether being controlled through the adaptive
turn-off control algorithm or other type of control algorithm,
however, there is need for improved control of synchronous
rectification switches in synchronous resonant switching
converters.
BRIEF SUMMARY
[0018] One aspect of the present disclosure is a control device for
a rectifier of a switching converter which avoids a reversal of a
current or "current inversion" through synchronous rectifier
transistors of the rectifier when the resonant converter
experiences a transient in the load current being supplied by the
resonant converter.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
[0019] For a better understanding of the present disclosure, a
preferred embodiment thereof is now described, purely by way of
non-limiting example and with reference to the annexed drawings,
wherein:
[0020] FIG. 1 is a diagram of an LLC-type resonant converter with a
center-tapped secondary winding and rectification of the output
current by means of diodes according to prior art;
[0021] FIG. 2 is a diagram of an LLC-type resonant converter with a
center-tapped secondary winding and rectification of the output
current by means of synchronous rectifier according to prior
art;
[0022] FIG. 3 shows the waveforms of the current flowing through a
transistor of the rectifiers of FIG. 2 and of its drain source
voltage;
[0023] FIG. 4 shows the waveforms of the current flowing through
one transistor of the rectifier in FIG. 2 with a premature turning
off of the transistor;
[0024] FIG. 5 shows a transistor of the rectifiers of FIG. 2 with
the parasitic elements, an RC compensation filter and a bypass
diode;
[0025] FIG. 6 is a block diagram of a switching converter
comprising a control device for a rectifier of a switching
converter according to a preferred embodiment of the present
disclosure;
[0026] FIG. 7 is a time diagram of the waveforms of the
drain-source voltages of the transistors of a synchronous rectifier
of prior art;
[0027] FIG. 8 is a schematic representation of the one of the
synchronous rectifier transistors in the rectifier of FIG. 6
including the stray inductance associated with the transistor;
[0028] FIG. 9 is signal timing diagram showing effect of the stray
inductance on the turn-off time of the synchronous rectifier
transistor of FIG. 8;
[0029] FIG. 10 is a functional block diagram illustrating the
operation of the adaptive turn-off control algorithm in generating
the gate drive signal control turn off of the synchronous rectifier
transistor of FIG. 8;
[0030] FIGS. 11A and 11B are signal timing diagrams that
graphically illustrate the operation of the adaptive turn-off
control algorithm as described with reference to FIGS. 8-10;
[0031] FIGS. 12A-12C are signal timing diagrams illustrating the
operation of the adaptive turn-off control algorithm over a number
of consecutive switching half-cycles;
[0032] FIGS. 13A and 13B illustrate current inversion of a
synchronous rectification current through a synchronous rectifier
transistor where a switching converter experiences a full-to-zero
load current transient;
[0033] FIG. 14 is a signal timing diagram showing plots of three
different values for a linear approximation of an anticipation time
along with the actual anticipation time as a function of switching
frequency of a switching converter;
[0034] FIG. 15 is a signal timing diagram of the difference between
the actual anticipation times of FIG. 14 and their linear
approximations as a function of the switching frequency;
[0035] FIGS. 16A-16C are signal timing diagrams illustrating how
the variation of the anticipation time resulting from a change in
the switching frequency results in current inversion in the context
of the adaptive turn-off control algorithm;
[0036] FIGS. 17A-17F are signal timing and functional diagrams
illustrating a current inversion prevention control algorithm for
controlling operation of the rectifier of the switching converter
of FIG. 6 to prevent current inversion even in the presence of load
current transients according to one embodiment of the present
disclosure;
[0037] FIG. 18 is a functional block diagram illustrating a
threshold calculation circuit of the current inversion prevention
control algorithm according to one embodiment of the present
disclosure; and
[0038] FIG. 19 is a functional block diagram one embodiment of the
comparator of FIG. 17 according to one embodiment of the present
disclosure.
DETAILED DESCRIPTION
[0039] Embodiments of the present disclosure are directed to a
current inversion prevention control algorithm for a rectifier of a
resonant converter, such as an LLC resonant converter, which avoids
a reversal of a current or "current inversion" through synchronous
rectifier transistors of the resonant converter when the resonant
converter experiences a high-to-low transient in a load current, as
will be described in more detail below. Embodiments of the present
disclosure may be utilized with the adaptive turn-off control
algorithm, but may also be utilized with other methods of
controlling the synchronous rectifier transistors of a resonant
converter. These embodiments utilize the actual drain-source
voltage Vdvs(t) across each synchronous rectifier transistor SR
taking into account the associated stray inductance L.sub.SR, as
discussed above, with this actual voltage Vdvs(t) being compared to
a threshold that is calculated cycle-by-cycle and is based on
sensed values of the voltage Vdvs(t) over the previous N-cycles, as
will be described in more detail below. This approach compensates
for variations in the value of the actual drain-source voltage
Vdvs(t) as a function of the switching frequency .omega..sub.sw of
the switching converter. Recall, the drain-source voltage
Vdvs(t)=(-Rdson.times.Isr-LSR.times..differential.Isr/dt) and thus
the component
( LSR .times. .differential. Isr .differential. t )
##EQU00003##
of the drain-source voltage Vdvs(t) caused by the stray inductance
L.sub.SR varies as a function of the switching frequency
.omega..sub.sw, which, in turn, causes the value of the
drain-source voltage Vdvs(t) to be a function of the switching
frequency, as will be described in more detail below. In the
equations contained herein, the stray inductance L.sub.SR may be
indicated as LSR.
[0040] As mentioned above, embodiments of the present disclosure
are not limited to being utilized with the adaptive turn-off
control algorithm. In the present disclosure, however, the
described embodiments will be described in conjunction with the
adaptive turn-off control algorithm to provide a better
understanding and example implementation of embodiments of the
present disclosure. Accordingly, a brief description of the
adaptive turn-off control algorithm will first be provided before
discussing embodiments of the present disclosure in more detail.
The adaptive turn-off control algorithm is described in detail in
U.S. Pat. No. 9,356,535 as mentioned above, with this patent being
incorporated herein in its entirety to the extent not inconsistent
with the specific teachings and definitions set forth herein.
[0041] Referring to FIG. 6, a control circuit 7 for a rectifier 5
of a switching converter 100 is shown according to an embodiment of
the present disclosure. The switching converter 100 includes power
switching circuitry including a power switching circuit block 1,
powered by a DC voltage and configured to generate a square wave
with a certain frequency under control of a first driver 20. For
example, the power switching circuit block 1 could have a
half-bridge or a full-bridge circuit topology (typically of MOSFET
transistors) but other power switching circuit blocks could be
equally adopted. The power switching circuitry of the switching
converter 100 further includes an impedance 30 connected to a
primary winding of a transformer 4, with the converter being
adapted to provide an output or load current Io/Iload to a load
(not shown in FIG. 6). In the embodiment of FIG. 6, the switching
converter 100 is an LLC resonant converter and the impedance 30 is
a resonant circuit. Although the control circuit 7 will be
described herein an executing the current inversion prevention
control algorithm in combination with an adaptive turn-off control
algorithm to control turn OFF of synchronous rectification
transistors SR1, SR2, the control circuit may also implement other
turn OFF control methods in combination with the current inversion
prevention control algorithm to control turn OFF of the transistors
SR1, SR2.
[0042] The square wave generated by the power switching circuit
block 1 is applied to the resonant circuit 30 tuned to a
fundamental frequency of the square wave. In this manner, because
of the selective frequency characteristics of the resonant circuit
30, the resonant circuit responds principally to this fundamental
frequency component and to a negligible degree to higher-order
harmonics. It follows that the circulating power can be modulated
by varying the frequency of the square wave, while maintaining the
duty cycle constant at 50%, and that, according to the
configuration of the resonant circuit 30, the currents and/or
voltages associated with the power flow will have a pattern that is
sinusoidal or sinusoidal at intervals.
[0043] The resonant circuit 30 is coupled to a transformer 4
including a primary winding L1 and center-tapped secondary windings
L2. A rectifier 5 comprising at least a transistor SR1, SR2 is
coupled to the center-tapped secondary winding L2 of the
transformer 4. Preferably, the rectifier 5 comprises a pair of
transistors SR1 and SR2 which have drain terminals connected with
respective terminals of two portions of the center-tapped secondary
winding L2 and source terminals connected with ground reference
GND. The center-tap CT of the secondary winding L2 is then
connected to a filter 6 which provides the output or load current
Io/Iload and output voltage Vo to a load (not shown) coupled to the
filter. The filter 6 could be a common parallel connection of a
capacitor and a resistance. The two transistors SR1, SR2 are
preferably MOSFET transistors with respective body diodes Dbr1,
Dbr2 and a suitably low on-resistance Ron, such that the
drain-source voltage drop across each transistor is significantly
lower than the voltage drop across a diode, like in the prior art
rectifier configuration shown in FIG. 1.
[0044] A control circuit 7 controls the rectifier 5, synchronously
driving the transistors SR1 and SR2 by means of two control signals
GD1 and GD2, applied to the control terminals of the transistors
SR1 and SR2, respectively, to control switching of the transistors
SR1 and SR2. The control circuit 7 receives as inputs the
drain-source voltages Vdvs1 and Vdvs2 of the transistors SR1, SR2,
and generates the two control signals GD1, GD2 for the transistors
SR1, SR2 based on the drain-source voltages Vdvs1, Vdvs2. The
currents ISR1, ISR2 are defined as being positive in the directions
shown in FIG. 6, namely flowing to output node OUTN of the
converter 100. A set of further signals and other temporal
parameters which are used by control circuit 7 are described
below.
[0045] The control circuit 7 includes a driving circuit 11 which
provides the control signals GD1, GD2 to the transistors SR1, SR2
and preferably includes a predictive comparator circuit 10. The
driving circuit 11 receives the output signals of a detecting
circuit 103 receiving the drain-source voltage Vdvs1 or Vdvs2 of
the transistors SR1, SR2.
[0046] The detecting circuit 103 is configured to detect when the
drain-source voltage Vdvs1, Vdvs2 crosses a voltage threshold
Vth_on and to provide an output signal ON_COMP responsive to the
drain-source voltage Vdvs1,Vdvs1 reaching the Vth_on threshold. The
voltage threshold Vth_on has a value that is related to the
threshold voltages of the body diodes Dbr1, Dbr2 of the transistors
SR1, SR2. In one embodiment, the detecting circuit 103 includes a
first comparator circuit configured to provide the output signal
ON_COMP as an output pulse signal starting in response to the
drain-source voltage Vdvs1, Vdvs2 becoming less than the fixed
voltage threshold Vth_on and ending in response to the drain-source
voltage Vdvs1, Vdvs2 increasing and reaching the same fixed voltage
threshold Vth_on. The detecting circuit 103 is also configured to
detect a zero-crossing event of the drain-source voltage Vdvs,
Vdvs2 of the transistors SR1, SR2 and to provide an output signal
ZCD in response to detecting such a zero-crossing event. To provide
the ZCD signal, in one embodiment the detecting circuit 103
includes a second comparator circuit configured to provide an
output pulse as the zero-crossing signal ZCD responsive to the
drain-source voltage Vdvs1, Vdvs2 crossing the zero voltage level,
namely the reference level ground GND in one embodiment.
[0047] A switching cycle SW of the switching converter 100 includes
two switching half-cycles SC1, SC2. Each switching half-cycle SC1,
SC2 of the switching converter 100 involves alternately one of the
transistors SR1, SR2, and in particular each half-cycle SC1, SC2
begins when the drain-source voltage Vdvs1, Vdvs2 is equal to a
threshold value VTH1 and ends when the drain-source voltage Vdvs1,
Vdvs2 reaches a threshold value VTH2, as seen in the signal diagram
of FIG. 7. Referring now to FIGS. 6 and 7, the transistor SR1 is
operative (i.e., is turned ON) during the first half-cycle SC1 of
the switching cycle SW while the transistor SR2 is operative (i.e.,
turned ON) during the second half-cycle SC2 of the switching cycle
SW. The drain-source voltages Vdvs1, Vdvs2 and the currents ISR1,
ISR2 of the MOSFET transistors SR1 and SR2 vary over time as shown
in FIG. 7 for these voltages and currents, which will be described
in more detail below with reference to FIGS. 11 and 12. The current
ISR2 has the same relative relationship to the voltage Vdvs2 during
half-cycle SC2 as does current ISR1 to voltage Vdvs1 during
half-cycle SC1. Each switching cycle SW includes the first and
second half-cycles SC1, SC2 of the converter 100, and thus includes
alternately turning ON and OFF the transistors SR1, SR2. Therefore,
each switching half-cycle SC1, SC2 of the converter 100 involves
activating one of the transistors SR1, SR2, and in particular each
half-cycle SC1, SC2 begins responsive to the drain-source voltage
Vdvs1, Vdvs2 becoming equal to the first threshold value VTH1 and
ends responsive to the same drain voltage Vdvs1, Vdvs2 reaching a
second threshold value Vth2. For example, in an application with a
switching cycle corresponding to a switching frequency of 100 KHz,
each of the half-cycles SC1, SC2 is equal to 5 microseconds (i.e.,
a switching cycle of 10 microseconds).
[0048] As already above-mentioned, the presence of parasitic
inductances Ldrain, Lsource and Ltrace, collectively referred to as
stray inductance L.sub.SR, determines an undesired earlier turn-off
T.sub.off of the transistors SR, SR2, which occurs at time t1 as
shown in FIG. 4 where the actual drain-source voltage Vdvs (i.e.,
Vdvs1 or Vdvs2) and the desired or ideal voltage Vdvs-ideal are
shown. This increases the residual conduction time T.sub.diode of
the body diode Dbr1,Dbr2, causing a loss of efficiency, which is
indicated as a duration LE in FIG. 4, due to the higher voltage
drop across the body diode Dbr1,Dbr2. For example, a typical
starting body diode residual conduction time T.sub.diode could be
on the order of 1 microsecond (i.e., the duration LE), while a
typical desired body diode residual conduction time value is on the
order of 60 nanoseconds for the ideal drain-source voltage
Vdvs-ideal.
[0049] In embodiments of the present disclosure, at the start of
the switching cycle SW, and particularly at the start of the
switching half-cycle SC1, SC2 where the transistor SR1, SR2 is
turned ON, as previously discussed with reference to FIG. 6, the
control circuit 7 according to a embodiments of the present
disclosure is configured to set a new off time instant T.sub.off1
of the transistor SR1, SR2 at the start switching half-cycle SC1,
SC2. The new off time instant T.sub.off1 is set so that the off
time instant immediately precedes a large time interval T1 with
respect the time instant of the zero crossing event of the
drain-source voltage Vdvs1, as described in more detail in the
previously incorporated U.S. Pat. No. 9,356,535. The new off time
instant T.sub.off1 has a fixed value, preferably a percentage of
the half-cycle SC1, SC2. This new off time instant T.sub.off1 is
appropriately chosen to take into account the gate drive discharge
time of the transistors SR1, SR2 and other additional factors
before a complete transistor turn OFF of the transistor SR1, SR2
occurs. A reasonable off time instant T.sub.off1 could be equal to
the 70% of the switching half-cycle SC1, SC2, but other suitable
percentages could also be used in order to set the off time instant
T.sub.off1.
[0050] While the adaptive turn-off control algorithm, and
alternative control algorithms for controlling rectification
circuits in synchronous switching converter, provide adequate
control of the switching converter in many applications, transients
in the load current Iload can result in inadequate control of the
converter. Transients in the load current Iload can occur, for
example, where a load is connected to or disconnected from the
switching converter during operation, such as where the switching
converter is part of a power supply for an electronic device such
as a laptop computer or smart phone. Large or high load current
transients result in variation in the regulated output voltage Vo,
which is then compensated for by control circuitry of the switching
converter to maintain the desired value of the output voltage Vo.
The control circuitry adjusts the switching frequency of the
switching converter in response to the transient in the load
current Iload and resulting change in output voltage Vo. As the
switching frequency is increased or decreased, depending on whether
the transient causes an increase or decrease in the load current
Iload, the adaptive turn-off control algorithm continues using the
last value of the programmable turn-off delay to turn OFF the
synchronous rectifier transistors SR.
[0051] This operation of the adaptive turn-off control algorithm in
response to transients in the load current Iload may result in the
synchronous rectifier transistors SR being turned OFF after the
zero current crossing of the current Isr, which may result in
"current inversion" in the switching converter. Current inversion
occurs where the synchronous rectifier current Isr through the
synchronous rectifier transistors SR reverses direction or
"inverts" before the synchronous rectifier controller turns OFF the
transistor SR. Current inversion is to be avoided because when the
current Isr reverses direction, namely becomes negative, and flows
in the direction opposite to the positive direction shown in FIG.
2, this negative current Isr flows from the rectifier output node
OUTN to ground GND and thereby discharges the rectifier output
node. This discharge of the output node OUTN wastes charge provided
to the rectifier output node, reducing the efficiency of the
converter. In addition to reducing the efficiency of the switching
converter, current inversion may also result in malfunction or
failure of the switching converter and should accordingly be
avoided for these additional reasons, as will be appreciated by
those skilled in the art.
[0052] In operation, the adaptive turn-off control algorithm takes
advantage of the anticipation effect of the stray inductance
L.sub.SR, namely the stray inductance Ltrace associated with the
printed circuit board (PCB) trace coupled to the transistor SR
along with the stray inductances Lsource, Ldrain of the package of
the transistor SR, as will now be discussed in more detail with
reference to FIGS. 8 and 9. FIG. 8 is a schematic representation of
the transistor SR and associated stray inductance L.sub.SR. The
ideal drain-to-source voltage Vdvs-ideal (see FIG. 4) is the
voltage at node d in FIG. 8, while the actual voltage
drain-to-source voltage Vdvs(t) is the voltage at node d', namely
the ideal voltage Vdvs-ideal (-R.sub.dson.times.I.sub.SR) plus the
voltage across the stray inductance L.sub.SR
(-LSR.times..differential.Isr/dt). In FIG. 9, the ideal voltage
Vdvs-ideal and the actual voltage Vdvs(t) are illustrated. The
vertical axis is voltage and the horizontal axis time in FIG. 9.
The effect of parasitic inductance L.sub.SR is illustrated in FIG.
9. The real or actual voltage Vdvs(t) is anticipated or leads in
time with respect to the ideal voltage Vdvs-ideal. This results in
the zero-crossing of the actual voltage Vdvs(t) occurring before
the zero-crossing of the ideal voltage Vdvs-ideal. In the context
of this description, the term "zero-crossing" is used to mean a
point indicating the voltage Vdvs-ideal or Vdvs(t) is approaching
zero and thus corresponds to threshold V.sub.TH_OFF as illustrated
by the horizontal dashed line in FIG. 9, and as previously
discussed with reference to FIG. 4. This anticipated or leading
characteristic of the actual voltage V.sub.DVS(t) is exploited in
the adaptive turn-off algorithm by means of a zero-crossing
comparator (see detector 103 of FIG. 6). In this way, the adaptive
turn-off control algorithm functions to turn-off the transistor SR
before the current I.sub.SR through the transistor reaches a zero
value or goes negative (i.e., before current inversion).
[0053] In operation, the adaptive turn-off control algorithm in
each switching half-cycle SC initially sets an estimated turn OFF
time Toff for the corresponding transistor SR that is active during
that half-cycle. The residual conduction time T.sub.diode of the
body diode Dbr of the transistor SR is then measured to determine
whether this conduction time is less than a time threshold
T.sub.th1. As long as the residual conduction time T.sub.diode of
the body diode Dbr is greater than the time threshold T.sub.th1,
the algorithm incrementally increases, cycle-by-cycle, a
programmable delay PD of the turn OFF time Toff of the transistor
SR and again measures the conduction time T.sub.diode of the body
diode Dbr of the transistor. As the programmable delay PD is
incrementally increased, the zero-crossing of the actual voltage
Vdvs(t) signal in FIG. 9 moves from the time t1 towards the time
t2, which corresponds to the zero-crossing of the ideal voltage
Vdvs-ideal. This movement of the zero-crossing point of the actual
voltage Vdvs(t) as the programmable delay PD is incrementally
increased is illustrated through the arrow 900 in FIG. 9. The
adaptive turn-off control algorithm thus incrementally increases
the programmable delay PD of the turn OFF time Toff for the
transistor SR until the measured conduction time T.sub.diode of the
body diode Dbr of the transistor is approximately equal to the
conduction time T.sub.diode-ideal of the body diode Dbr for the
ideal voltage Vdvs-ideal. Thus, the time threshold T.sub.th1
corresponds to the conduction time T.sub.diode-ideal, with the
programmable delay PD being incrementally adjusted to delay the
turn OFF of the transistor SR until the actual conduction time
T.sub.diode is approximately equal to the ideal conduction time
T.sub.diode-ideal, which corresponds to the zero-crossing of the
actual voltage Vdvs(t) occurring at the time t2 in FIG. 9. In this
way, the adaptive turn-off control algorithm adjusts the actual
turn-off time Toff of the transistor SR by a stray inductance time
T.sub.LSTRAY, which as shown in FIG. 9 corresponds to the time
interval from time t1 to time t2.
[0054] This operation of the adaptive turn-off control algorithm as
just described with reference to FIGS. 8 and 9 is represented in
the functional diagram of FIG. 10. The actual voltage Vdvs(t) is
compared to the threshold VTH-OFF, which is represented in the
figure as a comparator, and upon detecting Vdvs(t)=VTH-OFF an
initial gate drive signal GD' is generated. This initial gate drive
signal GD' is then delayed with the programmable delay PD as
previously described to generate a delayed gate drive signal GD''
that is then provided through a driver DRV as the actual gate drive
signal GD that is provided to control switching of the synchronous
rectifier transistor SR. The programmable delay PD is incrementally
adjusted, namely increased or decreased, to delay the turn OFF of
the transistor SR until the actual conduction time T.sub.diode of
the body diode Dbr of the transistor SR is approximately equal to
the ideal conduction time T.sub.diode-ideal that the body diode Dbr
would have in the absence of the stray inductance L.sub.SR
(L.sub.SR=Lsource+Ldrain+Ltrace), as described above.
[0055] FIGS. 11A and 11B are signal timing diagrams that
graphically illustrate the operation of the adaptive turn-off
control algorithm as just described with reference to FIGS. 8-10.
The voltage Vdvs and current ISR through one of the synchronous
rectification transistors SR during a half-cycle SC during which
that transistor is turned ON and then OFF are shown in FIG. 11A.
FIG. 11B shows the gate drive signal GD that is applied by the
control circuit 7 (FIG. 6) to turn ON the transistor SR at a time
t1' to start the half-cycle SC, and to turn OFF of the transistor
initially at time t1 and thereafter at times that are incrementally
delayed relative to time t1 until the turn OFF of the transistor SR
occurs at time t2. The time t2 corresponds to the desired turn OFF
time of the transistor SR, meaning this is the time that makes the
conduction time T.sub.diode of the body diode Dbr of the transistor
SR equal to the ideal conduction time T.sub.diode-ideal, as
indicated in FIG. 11A. In FIG. 11A, the vertical axis represents
voltage or current (V/I), namely voltage Vdvs and current I.sub.SR,
while the horizontal axis represents time t. Similarly, in FIG. 11B
the vertical axis represents voltage V while the horizontal axis
represents time t.
[0056] As seen in FIG. 11B, the gate drive signal GD initially goes
low at time t1 to turn OFF the transistor SR. The adaptive turn-off
control algorithm then measures the conduction time of body diode
Dbr by comparing the voltage Vdvs(t) to the threshold V.sub.TH-OFF
as described in more detail in the U.S. Pat. No. 9,356,535. When
the algorithm measures the conduction time T.sub.diode after the
initial turn OFF at time t1, the algorithm determines whether the
conduction time T.sub.diode is greater than the time
T.sub.diode-ideal (T.sub.diode>T.sub.diode-ideal) and if so the
algorithm adds an incremental delay T.sub.step to the initial turn
OFF time at time t1, so the new turn-off time
Toff=(Toff1+T.sub.step). The algorithm continues incrementing the
turn-off time Toff by the incremental delay T.sub.step each
half-cycle SC of the switching converter 100. The algorithm does so
for each half-cycle SC1, SC2, with the transistor SR1 being
controlled, measured, and the turn-off time Toff adjusted in the
first half-cycle SC1 and the transistor SR1 being controlled,
measured, and the turn-off time Toff adjusted in the second
half-cycle SC2. The present description describes the operation of
the adaptive turn-off control algorithm for one half-cycle SC and
applies to the control of each of the transistors SR1 and SR2
during the half-cycles SC and SC2, respectively.
[0057] FIG. 11B illustrates the operation of the adaptive turn-off
control algorithm in incrementally delaying the gate drive signal
GD from the time t1 to the time t2 at which the conduction time
T.sub.diode is approximately equal to the ideal conduction time
T.sub.diode-ideal (T.sub.diode=T.sub.diode-ideal). Each vertical
dashed line represents the turn-off time Toff at the respective
time steps T.sub.step added to the prior turn-off time Toff, with
the algorithm adding an incremental delay D.sub.step each
half-cycle SC to provide the new turn-off time Toff. In other
words, the programmable delay PD (FIG. 10) is incrementally
adjusted by the incremental delay D.sub.step each half-cycle SC to
decrease the conduction time T.sub.diode. This programmable delay
PD is represented in FIG. 11B as an arrow PD pointing to the right
between the times t1 and t2. The decreasing of the conduction time
T.sub.diode of the body diode Dbr of the transistor SR to minimize
the conduction time T.sub.diode until
(T.sub.diode=T.sub.diode-ideal), as illustrated in FIG. 11B as just
described, also corresponds to maximizing the conduction time of
the transistor SR each half-cycle SC.
[0058] The description of the adaptive turn-off control algorithm
set forth above with reference to FIGS. 8-11 ignores a gate
discharge time of the transistor SR, along with additional factors
associated with the actual operation of the transistor. Briefly,
each transistor SR takes a finite time to turn OFF and thus does
not completely turn OFF immediately in response to the gate drive
signal GD. Thus, the actual deactivation of the gate drive signal
GD occurs in advance of the desired turn OFF time of the transistor
SR to compensate for the gate discharge time and other factors to
ensure the transistor is turned completely OFF at the desired
turn-off time, which is also described in more detail in the U.S.
Pat. No. 9,356,535.
[0059] FIGS. 12A-12C are signal timing diagrams illustrating the
operation of the adaptive turn-off control algorithm over a number
of consecutive switching half-cycles SC. In each of these figures,
the horizontal axis is time t in the vertical axis represents
current I and voltage V, with the current I.sub.SR and voltage Vdvs
signals being shown in each signal diagram. The switching
half-cycle SC of each is labeled separately, with an initial
half-cycle CYCLE[0] shown in FIG. 12A, a next half-cycle CYCLE[1]
shown in FIG. 12B, and an nth half-cycle CYCLE[n] shown in FIG.
12C. The ideal voltage V.sub.dvs-ideal of the transistor SR being
controlled is illustrated as a dashed line in each of the FIGS.
12A-12C. During the half-cycle CYCLE[0] in FIG. 12A, a
zero-crossing of the actual voltage Vdvs(t) is detected and in
response thereto a zero-crossing detection signal ZCD is activated
at a time t1. In response to the ZCD signal going active at time
t1, the control circuit 7 (FIG. 6) deactivates the gate drive
signal GD of the corresponding transistor SR being controlled
during the half-cycle CYCLE[0], which is also shown in the figure.
The situation in the half-cycle CYCLE[0] corresponds to the initial
operation of the control circuit 7 in controlling the transistor SR
in the rectifier 5 (FIG. 6), namely before the adaptive turn-off
control algorithm has begun delaying the gate drive signal GD to
start reducing the conduction time T.sub.diode of the body diode
Dbr of the transistor SR.
[0060] In FIG. 12B, during the half-cycle CYCLE[1] the adaptive
turn-off control algorithm delays the gate drive signal GD by a
first delay T.sub.DEL_OFF relative to the gate drive signal in the
cycle CYCLE[0]. The delay T.sub.DEL_OFF is accordingly a first
incremental or step delay added to the gate drive signal GD to
reduce the conduction time T.sub.diode of the body diode Dbr of the
transistor SR as discussed above. Thus, during this initial
half-cycle CYCLE[0] the programmable delay PD=0 and in the second
half-cycle CYCLE[1] the programmable delay PD=T.sub.DEL_OFF, where
this incremental delay is labelled T.sub.DEL_OFF1 in FIG. 12B. The
adaptive turn-off control algorithm continues operating in this
manner each of the next half-cycles CYCLE[2]-CYCLE[n-1], namely
adding the incremental delay T.sub.DEL_OFF to the programmable
delay PD of the gate drive signal GD each half-cycle and measuring
the conduction time T.sub.diode of the body diode Dbr of the
transistor SR. Finally, in the nth half-cycle CYCLE[n], the
programmable delay
PD=T.sub.DEL_OFF1+T.sub.DEL_OFF2+T.sub.DEL_OFF3+T.sub.DEL_OFF4 . .
. +T.sub.DEL_OFFn, where the delay T.sub.DEL_OFF2 is the
incremental delay added during the half-cycle CYCLE[1],
T.sub.DEL_OFF3 is the incremental delay added during the half-cycle
CYCLE[2], delay T.sub.DEL_OFF4 is the incremental delay added
during the half-cycle CYCLE[3], and delay T.sub.DEL_OFFn is the
incremental delay added during the half-cycle CYCLE[n]. In the
cycle CYCLE[n], the conduction time T.sub.diode of the body diode
Dbr of the transistor SR is approximately equal the ideal
conduction time T.sub.diode-ideal and thus the gate drive signal GD
is operated with this programmable delay PD in cycle CYCLE[n] and
subsequent cycles to control the deactivation of the transistor SR.
In this way, the adaptive turn-off control algorithm accounts for
the stray inductance L.sub.SR associated with the transistor SR to
turn OFF the transistor at the proper time each half-cycle to
minimize the conduction time to T.sub.diode of the body diode Dbr
and improve the efficiency of the switching converter 100. As the
actual conduction time T.sub.diode approaches the ideal conduction
time T.sub.diode-ideal, the adaptive turn-off control algorithm may
increment, or decrement, the programmable delay PD by a smaller or
finer incremental delay than the incremental delay T.sub.DEL_OFF
initially utilized to increment this programmable delay in order to
provide a final programmable delay PD that results in an actual
conduction time T.sub.diode that is more closely equal to the ideal
conduction time T.sub.diode-ideal, as described in more detail in
the U.S. Pat. No. 9,356,535.
[0061] This operation of the adaptive turn-off control algorithm as
described above assumes that the anticipatory effects of the stray
inductance L.sub.STRAY=L.sub.SR is constant with respect to the
switching frequency .omega..sub.sw of the switching converter 100.
Under this assumption, however, after a transient in the load
current I.sub.O/I.sub.LOAD of the switching converter 100, the
output voltage V.sub.O of the converter decreases or increases,
resulting in a primary control loop of the switching converter
driving the resonant circuit 30 (FIG. 6) at a new switching
frequency .omega..sub.SW. Although not described in detail herein,
the primary control loop of the resonant switching converter 100
compares the regulated output voltage V.sub.O to a desired value
and adjusts the switching frequency .omega..sub.SW of the switching
converter to maintain the regulated output voltage at this desired
value, as will be understood by those skilled in the art. This new
switching frequency .omega..sub.SW of the switching converter 100
results in a different anticipatory effect of the stray inductance
L.sub.SR on the voltage Vdvs(t) across the synchronous
rectification transistors SR. While this is true, after such a
transient in the load current I.sub.O/I.sub.LOAD and subsequent
variation in the load switching frequency .omega..sub.SW of the
switching converter 100, the adaptive turn-off control algorithm
continues using the last value of programmable delay PD determined
prior to the transient in the load current.
[0062] Typically, this operation results in the adaptive turn-off
control algorithm continuing to turn OFF the transistor SR earlier
than the zero-crossing of the current I.sub.SR through the
transistor SR and avoiding current inversion, which was described
in more detail above. Where the switching converter 100 experiences
a high-to-low transient in the load current I.sub.O/I.sub.LOAD,
however, the regulated output voltage V.sub.O can experience some
"overshoot," meaning the regulated output voltage becomes greater
than the desired regulated output voltage. As a result, the primary
control loop increases the switching frequency .omega..sub.SW to
decrease the regulated output voltage V.sub.O towards the desired
value.
[0063] In a first approximation, the anticipation effect of the
stray inductance L.sub.SR can be considered a constant parameter,
but in some specific cases this approximation is insufficient to
describe the real phenomenon of this stray inductance. In these
cases, after a transient from a maximum or full load current
I.sub.O/I.sub.LOAD to a zero load current, the regulated output
voltage V.sub.O increases beyond the desired value because the
primary control loop is not fast enough to control the switching
converter to balance this transient. As a result, the primary
control loop increases the switching frequency .omega..sub.SW to
compensate for this variation in the regulated output voltage
V.sub.O and the anticipation effect of the stray inductance
L.sub.SR is reduced. This means that if the adaptive turn-off
control algorithm goes on exploiting the last adaptive turn-off
delay value, namely the last value of the programmable delay PD,
the transistor SR is turned OFF later which may result in current
inversion of the current I.sub.SR.
[0064] FIGS. 13A and 13B illustrate the current inversion of the
current I.sub.SR that can occur in the situation described above
for a transient from a maximum or full load current
I.sub.O/I.sub.LOAD to a zero load current, which will be referred
to as a "full-to-zero load current transient" in the present
description. FIG. 13A illustrates operation of the switching
converter 100 prior to the full-to-zero load current transient and
corresponds to the operating condition described above for the
cycle CYCLE[n] in FIG. 12C and subsequent half-cycles once the
programmable delay PD (PD=T.sub.DEL_OFF+T.sub.DEL_OFF2 . . .
+T.sub.DEL_OFFn) has been adjusted such that the conduction time
T.sub.diode=T.sub.diode-ideal.
[0065] FIG. 13B illustrates operation in a half-cycle CYCLE[n+1]
after a full-to-zero load current transient has occurred and the
switching frequency .omega..sub.SW is increased to compensate for
the overshoot of the regulated output voltage V.sub.O. As seen in
FIG. 13B, the anticipatory effect of the stray inductance L.sub.SR
is reduced in the cycle CYCLE[n+1], resulting in the zero-crossing
of the voltage Vdvs(t) occurring at time t1' that is later in time
than the time t1. The actual voltage Vdvs(t) in FIG. 13B is delayed
relative to FIG. 13A (i.e., the anticipatory effect of the stray
inductance L.sub.SR is less) and the resulting Vdvs(t) voltage is
closer to the ideal voltage Vdvs-ideal shown in the dashed line. As
a result, when the zero-crossing of the voltage Vdvs(t) is detected
at time t1' and the deactivation of the gate drive signal GD
delayed by the programmable delay PD from this point, the gate
drive signal does not deactivate the transistor SR until time t2'
which is later in time relative to time t2. During the programmable
delay PD in FIG. 13B (i.e., from the time t1' to the time t2' in
FIG. 13B) the transistor SR is still turned ON and current
inversion of the current I.sub.SR occurs. This current inversion of
the current I.sub.SR is labelled in FIG. 13B and corresponds to the
current becoming negative and thus flowing from the output node
OUTN (FIG. 6) to ground and thus undesirably discharge the output
node OUTN, as previously mentioned above with reference to FIG.
6.
[0066] In order to better describe this phenomenon, a closed-form
expression for the anticipation effect of the stray inductance
L.sub.SR is required. To obtain such a closed-form expression,
assume the current I.sub.SR has an almost sinusoidal shape during
the last part of the switching cycle according to the following
equation:
I.sub.SR(t)=I.sub.PK sin(.omega..sub.SWt) Eqn. 1
Under these conditions, the actual or real voltage V.sub.DVS(t) is
given by the following equation:
V DVS ( t ) = R DS ( on ) I SR ( t ) + L SR .differential. I SR ( t
) .differential. t = R DS ( on ) I PK sin ( .omega. SW t ) + L SR
.omega. SW I PK cos ( .omega. SW t ) Eqn . 2 ##EQU00004##
This expression can be rewritten splitting the time component and
the phase shift component as follows:
##STR00001##
where T.sub.L.sub.SR is an anticipation time caused by the stray
inductance L.sub.SR. The anticipation time T.sub.L.sub.SR of Eqn. 3
is referred as T.sub.LSR in the following description (i.e.,
T.sub.L.sub.SR=T.sub.LSR). Equating Eqns. (2) and (3) it is
possible to obtain the following expression for the phase
shift:
{ K sin ( .PHI. ) = L SR .omega. SW I PK K cos ( .PHI. ) = R DS (
on ) I PK .fwdarw. .PHI. = arctan ( L SR R DS ( on ) .omega. SW )
Eqn . 4 T L SR = .PHI. / .omega. SW = 1 .omega. SW arctan ( L SR R
DS ( on ) .omega. SW ) .apprxeq. L SR R DS ( on ) Eqn . 5
##EQU00005##
The resistance resistance R.sub.DS(on) is the ON resistance of the
synchronous rectifier transistor SR, and is also referred to as
R.sub.dson in the above description, as previously mentioned.
[0067] The last approximation in Eqn. (5) is for the anticipation
time T.sub.LSR and is true only when
L SR R DS ( on ) << 1 / .omega. SW , ##EQU00006##
and in this situation the anticipation time T.sub.LSR is constant
as a function of the switching frequency
f.sub.SW(.omega..sub.SW=2.pi.f.sub.SW). This is a linear
approximation of the arctan function and applies where the argument
(L.sub.SR/R.sub.DS(on).omega..sub.SW) is much less than one
((L.sub.SR/R.sub.DS(on).omega..sub.SW)<<1). FIG. 14 is a
graph showing plots of three different values for the linear
approximation of the anticipation time T.sub.LSR along with the
anticipation time T.sub.LSR as given by Eqn. 5 as a function of
switching frequency f.sub.sw for these three values of this
approximation. In FIG. 14, the line 1400 is the linear
approximation of the anticipation time T.sub.LSR where the stray
inductance L.sub.SR=4 nH, and the ON resistance R.sub.DS(on)=4
m.OMEGA. so that T.sub.LSR=(4 nH/4 m.OMEGA.)=1 .mu.s=1000 ns. The
line 1402 shows the plot of Eqn. 5 where L.sub.SR/R.sub.DS(on)=1000
ns. Similarly, the line 1404 is the linear approximation of the
anticipation time T.sub.LSR where the stray inductance L.sub.SR=8
nH, and the ON resistance R.sub.DS(on)=4 m.OMEGA. so that
anticipation time T.sub.LSR=(8 nH/4 m.OMEGA.)=2 .mu.s=2000 ns and
the line 1406 shows the plot of Eqn. 5 where
L.sub.SR/R.sub.DS(on)=2000 ns. Finally, the line 1408 is the linear
approximation of the anticipation time T.sub.LSR where the stray
inductance L.sub.SR=16 nH, and the ON resistance R.sub.DS(on)=4
m.OMEGA. so that anticipation time T.sub.LSR=(16 nH/4 m.OMEGA.)=4
.mu.s=4000 ns and the line 1410 shows the plot of Eqn. 5 where
L.sub.SR/R.sub.DS(on)=4000 ns.
[0068] FIG. 14 shows how the actual or real anticipation time
T.sub.LSR varies as a function of switching frequency f.sub.SW. The
adaptive turn-off control algorithm assumes the linear
approximation for the anticipation time T.sub.LSR and accordingly
assumes the value of the anticipation time does not vary as a
function of switching frequency f.sub.SW, which is not accurate
over the range of switching frequencies shown in FIG. 14. After a
high-to-low transient in the load current Iload, the primary
control loop of the resonant switching converter 100 (FIG. 6)
increases the switching frequency f.sub.SW to compensate for
variation in the output voltage Vout to thereby provide desired
regulation of this output voltage. This variation
(.DELTA.f.sub.SW/f.sub.SW) in the switching frequency f.sub.SW
could be on the order of 10%-15%, which causes a change or
variation .DELTA.T of the anticipation time T.sub.LSR on the order
of 50 ns, which is comparable to turn-off time of the transistor SR
or the target body diode conduction time T.sub.diode-ideal. In this
situation, namely a high-to-low load current transient, the
variation .DELTA.T is more specifically a reduction in the
anticipation time T.sub.LSR. If the variation .DELTA.T of the
anticipation time T.sub.LSR reaches or exceeds the target or
desired conduction time T.sub.diode-ideal (e.g., 50 ns) of the body
diode Dbr, current inversion occurs where the synchronous rectifier
current Isr through the transistor SR reverses direction, as
previously described.
[0069] FIG. 15 is a graph of the variation .DELTA.T of the
anticipation time T.sub.LSR as a function of the switching
frequency f.sub.SW. The variation .DELTA.T in the anticipation time
T.sub.LSR represents the error or variation between the ideal
constant linear approximation value of the anticipation time
T.sub.LSR and the real or actual value of the anticipation time as
a function of switching frequency f.sub.SW. The line 1500 shows the
variation .DELTA.T of the anticipation time T.sub.LSR from the
linear approximation value of T.sub.LSR=(4 nH/4 m.OMEGA.)=1
.mu.s=1000 ns as a function of switching frequency f.sub.SW. The
line 1502 shows the variation .DELTA.T of the anticipation time
T.sub.LSR from the linear approximation value of T.sub.LSR=(8 nH/4
m.OMEGA.)=2 .mu.s=2000 ns as a function of switching frequency
f.sub.SW while the line 1504 shows the variation .DELTA.T of the
anticipation time T.sub.LSR from the linear approximation value of
T.sub.LSR=(16 nH/4 m.OMEGA.)=4 .mu.s=4000 ns as a function of
switching frequency.
[0070] FIGS. 16A-16C are signal timing diagrams illustrating how
the variation .DELTA.T of the anticipation time T.sub.LSR resulting
from a change in the switching frequency f.sub.SW from a half-cycle
n to the next half-cycle n+1 results in current inversion in the
context of the adaptive turn-off control algorithm. In the example
of FIGS. 16A-16C, the switching converter 100 experiences a
high-to-low current transient. If the variation .DELTA.T of the
anticipation time T.sub.LSR is greater than a target turn-off time
Toff according to the adaptive turn-off algorithm, a current
inversion of the current I.sub.SR caused by the delayed turn-off
Toff provided by the adaptive turn-off control algorithm can occur.
In FIG. 16A, the vertical axis is the voltage of the Vdvs signal
and the horizontal axis is time t. In FIGS. 16B and 16C the
vertical axis is the voltage of the output signal ZCD indicating a
zero-crossing event of the drain-source voltage Vdvs of the
transistor SR (FIG. 6) and the horizontal axes are time t.
[0071] FIGS. 16A-16C show, for a first half-cycle n, the ideal
voltage signal Vdvs-ideal 1600 and the real or actual signal Vdvs
1602 for this first half-cycle n. The cycle n is the cycle before
the load transient occurs. Similarly, in a second half-cycle n+1
the ideal voltage signal Vdvs-ideal 1604 and the real or actual
signal Vdvs 1606 for this second half-cycle are shown, where the
load transient occurs between the cycle n and cycle n+1. The ideal
signals Vdvs 1600, 1604 are not affected by the stray inductance
L.sub.SR and thus these signals are directly proportional to the
current I.sub.SR. In contrast, the real or actual signals Vdvs
1602, 1606 are affected by the stray inductance L.sub.SR and the
variation .DELTA.T of the anticipation time T.sub.LSR due to the
change in switching frequency f.sub.SW from the half-cycle n to the
half-cycle n+1.
[0072] FIG. 16A shows these various ideal and actual signals DVS
during the half-cycles n and n+1, with the time t along the
horizontal axis being expanded or zoomed-in near the zero crossing
of the load current Iload associated with the signals in the
figure. The load current Iload (FIG. 6) is equal to the currents
(ISR1+ISR2), where one of these currents ISR1, ISR2 is equal to
zero each half-cycle SC so that the load current Iload is equal to
either ISR1 or ISR2 each half-cycle. FIG. 14 illustrates the effect
of the variation .DELTA.T of the anticipation time T.sub.LSR
between the half-cycle n before the load transient and cycle n+1
after the load transient. The anticipation time T.sub.LSR of the
half-cycle n is designated T.sub.LSR [n] and the anticipation time
T.sub.LSR of the half-cycle n+1 is designated T.sub.LSR[n+1] in the
figure.
[0073] As seen in FIGS. 16A-16C, the anticipation time
T.sub.LSR[n+1] (FIG. 16C) is less than the anticipation time
T.sub.LSR[n] (FIG. 16B) due to the increase of the switching
frequency f.sub.SW from the cycle n to the cycle n+1. The
zero-crossing of the voltage Vdvs occurs at a time t1 for each of
the half-cycles n and n+1. In the half-cycle n, the programmable
delay PD provided by the adaptive turn-off control algorithm is a
time starting from the detection of the zero-crossing of the
voltage Vdvs at time t1 until the turn-off time Toff at a time t2.
The duration of this programmable delay PD is also designated as
the period T.sub.DEL_OFF from time t1 to t2 in the figures. As seen
in FIGS. 16A and 16B, in the half-cycle n, prior to the load
transient, the programmable delay PD provided by the adaptive
turn-off control algorithm turns OFF the transistor SR at the time
Toff=t2 in the figure, providing the ideal conduction time
T.sub.diode-ideal for the body diode Dbr of the transistor.
[0074] Referring to FIG. 16C, in the half-cycle n+1 after the
occurrence of the load transient, the adaptive turn-off control
algorithm again provides the programmable delay PD to turn OFF the
transistor SR at the time Toff=t2. While this is true, the change
in switching frequency f.sub.SW due to the load transient and
resulting change in the anticipation time T.sub.LSR[n+1] means the
zero-crossing of the ideal signal Vdvs-ideal occurs earlier as seen
in the figures, and the transistor SR should be turned OFF earlier
at time T'.sub.off, which is at a time t2' in FIGS. 16A-16C. The
anticipation time T.sub.LSR[n+1] corresponds to the time from time
t1-t2'. The adaptive turn-off control algorithm does not, however,
turn OFF the transistor SR at time the T'.sub.off but again turns
OFF the transistor at time Toff. As a result, the zero-crossing of
signal Vdvs occurs at time t2' in half-cycle n+1 but the transistor
SR is not turned OFF until time t2, resulting in current inversion
of the current I.sub.SR (i.e., I.sub.SR goes negative) prior to the
transistor being turned OFF. This current inversion is undesirable
as previously discussed.
[0075] As seen from this description of the adaptive turn-off
control algorithm with reference to FIGS. 16A-16C, the programmable
delay PD evaluated during the half-cycle n is used for half-cycle
n+1. This programmable delay PD is updated only at the end of
half-cycle n+1 and this update delay then used in the next
half-cycle n+2. When the anticipation time T.sub.LSR is constant no
problem in the operation of the adaptive turn-off control algorithm
results. When there is a variation of the anticipation time
T.sub.LSR due to a high-to-low load current transient as described
with reference to FIGS. 16A-16C, however, due to the variation of
the switching frequency f.sub.SW a current inversion can occur as
just described. As illustrated in FIGS. 16B and 16C, in this
situation T.sub.LSR[n+I]T.sub.LSR[n] and this reduction in the
anticipation time .DELTA.T (i.e., from t3 to t2') results in a
delay turning OFF of the transistor SR, causing the unwanted
current inversion of the current I.sub.SR.
[0076] FIGS. 17A-17F are signal timing and functional diagrams
illustrating operation of a current inversion prevention control
algorithm for controlling operation of the rectifier 5 of the
switching converter 100 of FIG. 6 to prevent current inversion even
in the presence of load current transients according to one
embodiment of the present disclosure. FIG. 17A is signal timing
diagram showing the drain-to-source voltage signal Vdvs of the
activated transistor SR1, SR2. More specifically, a first
half-cycle CYCLE[n] is shown in which the transistor SR1 is
activated (i.e., turned ON). The current and voltage associated
with the transistor SR1 are labelled I.sub.SR1 and V.sub.DVS1,
respectively. A second half-cycle CYCLE[n+1] is shown in which the
transistor SR2 is activated, with the current and voltage
associated with the transistor SR2 being labelled I.sub.SR2 and
V.sub.DVS2, respectively. Finally, a third half-cycle CYCLE[n+2] is
shown in which the transistor SR1 is once again activated and the
current I.sub.SR1 and voltage V.sub.DVS1 shown for this
half-cycle.
[0077] FIGS. 17B and 17C are signal timing diagrams showing the
gate drive signals GD1, GD2 applied to control the activation and
deactivation (i.e., turning ON and OFF) of the transistors SR1 and
SR2, respectively, as shown in FIG. 6. FIG. 17D shows the magnitude
of the voltage Vdvs across the activated transistor SR1 or SR2,
which is compared to a threshold TH=(.alpha.V.sub.DVS_PK) in each
of the half-cycles CYCLE[n], CYCLE[n+1] and CYCLE[n+2], which is
represented through a comparator 1700 shown in FIG. 17F, as will be
described in more detail below. The parameter a is a threshold
adjustment factor that will be discussed in more detail below. FIG.
17E shows an output signal COMP generated by the comparator 1700,
as will also be described in more detail below. Each of the FIGS.
17A-17E shows voltage, or voltage and current, along the vertical
axis and time t along the horizontal axis.
[0078] The control circuit 7 of FIG. 6 is configured to execute
this current inversion prevention control algorithm, where this
control circuit may be formed through suitable circuitry that will
be understood by one skilled in the art, such as hardware, software
or firmware executing on suitable processing circuitry, or
combinations of these types of circuits. In operation, the current
inversion prevention control algorithm detects the voltage signals
Vdvs of the transistor SR and controls the transistor to avoid
current inversion of the current I.sub.SR before the transistor is
turned OFF each half-cycle, as will now be explained in more detail
with reference to FIGS. 17-19. As previously mentioned, the current
inversion prevention control algorithm is described in the present
application in the context of or as being used with the adaptive
turn-off control algorithm, but may also be utilized with other
control algorithms utilized to control the transistors SR in a
rectification circuit of a switching converter.
[0079] In operation, the current inversion prevention control
algorithm compares the actual voltage signal Vdvs sensed each
half-cycle SC with a peak voltage reference threshold TH that is
calculated cycle-by-cycle from the sensed voltage signal Vdvs each
half-cycle. Cycle-by-cycle means the voltage signal Vdvs of the
transistor SR1, SR2 being activated each half-cycle SC is sensed,
and it is utilized in calculating the threshold TH. The result of
this comparison is then used each half-cycle SC to prevent current
inversion of the current I.sub.SR if a high-to-low load current
transient occurs, with this comparison being represented as the
comparator 1700 in FIG. 17, as will now be explained in more
detail. The peak voltage reference threshold TH actually includes a
first peak voltage reference threshold for the voltage signal Vdvs
of the transistor SR1 and a second peak voltage reference threshold
for the transistor SR2, as will be described in more detail below
with reference to FIG. 18.
[0080] The current inversion prevention control algorithm evaluates
the average voltage V.sub.DVS_AVG of the voltage signal Vdvs during
a conduction time TSW of the transistor SR over the previous N
half-cycles SC. This average voltage signal V.sub.DVS_AVG of the
signal Vdvs is not affected by the stray inductance L.sub.SR and is
proportional to the load current Iload. Recall, the load current
Iload=ISR1+ISR2 (see FIG. 6). If the current I.sub.SR is
approximated to have an almost sinusoidal shape during the last
part of the switching half-cycle SC, the average V.sub.DVS_AVG and
the peak V.sub.DVS_PK of the voltage signal Vdvs are related
according through the following equation:
V DVS _ PK = .pi. 2 V DVS _ AVG Eqn . 6 ##EQU00007##
This means that it is possible to calculate the peak value
V.sub.DVS_PK of the Vdvs signal from the average value
V.sub.DVS_AVG. Thus, the current inversion prevention control
algorithm determines the peak voltage reference threshold TH based
on the peak value V.sub.DVS_PK, with the threshold TH, for example,
being set equal to 75% of the peak value V.sub.DVS_PK. Thus, the
peak voltage reference threshold TH=(.alpha.V.sub.DVS_PK), where a
is a threshold adjustment factor and .alpha.=0.75 in the example
being described. The current inversion prevention control algorithm
then compares this peak voltage reference threshold TH to the
actual voltage signal Vdvs during the conduction time TSW of the
transistor SR.
[0081] In operation, during half-cycles SC in which no load current
transient occurs, the comparator 1700 generates the output signal
COMP that goes active within a certain percentage of the conduction
time TSW of the transistor SR during that half-cycle SC. In the
example of FIGS. 17A-17E, the comparator 1700 activates the COMP
signal within, for example, 75% of the conduction time TSW when no
load current transient has occurred. The conduction time TSW is
multiplied by a conduction time adjustment factor R to define this
certain percentage of the conduction time, where R=0.75 in the
example embodiment being described. The COMP signal generated by
comparator 1700 is shown in FIG. 17E. During the first half-cycle
CYCLE[n] the comparator 1700 compares the voltage signal V.sub.DVS1
of the activated transistor SR1 to the threshold
TH=(.alpha.V.sub.DVS_PK). This threshold TH is related to average
value V.sub.DVS_AVG of the voltage signal Vdvs through Eqn. 6 as
previously described, with the average value V.sub.DVS_AVG being
shown in FIG. 17D. No load current transient occurs in CYCLE[n] and
thus at a time t1 the comparator 1700 detects the voltage
V.sub.DVS1 has reached the threshold TH and activates the COMP
signal. The same operation occurs in the second illustrated
half-cycle CYCLE[n+1], with the comparator 1700 comparing the
voltage signal V.sub.DVS2 of the activated transistor SR2 to the
threshold TH=(.alpha.V.sub.DVS_PK). At time t2, the comparator 1700
detects the voltage signal V.sub.DVS2 has reached the threshold TH
and activates the COMP signal.
[0082] This operation of the comparator 1700 in generating the COMP
signal in CYCLE[n] and CYCLE[n+1} illustrates that in half-cycles
SC where no load current transient has occurred, the sensed voltage
signal V.sub.DVS1,V.sub.DVS2 reaches the threshold within a certain
percentage (as defined by adjustment parameter p) of the conduction
time TSW of the activated transistor SR1, SR2. Conversely, where a
load current transient occurs the comparator 1700 does not activate
the COMP signal within this certain percentage of the conduction
time TSW, as will now be explained in more detail with reference to
CYCLE[n+2] illustrated in FIGS. 17A-17F. In this example, a
high-to-low load current transient, such as from maximum load
current to zero load current Iload, occurs between the CYCLE[n+1]
and the CYCLE[n+2], as represented by arrow 1702 in FIG. 17A. As a
result of this load current transient, the voltage V.sub.DVS1
sensed by the comparator 1700 in the cycle CYCLE[n+2] does not
reach the threshold TH=(.alpha.V.sub.DVS_PK) within the certain
percentage of the conduction time TSW. In fact, in the illustrated
example the voltage V.sub.DVS1 does not reach the threshold TH at
all in CYCLE [n+2]. The comparator 1700 therefore does not activate
the COMP signal in CYCLE[n+2], which can be used to indicate the
detection of a load current transient and corresponding adjustment
to the control of the switching converter 100 (FIG. 6) to prevent
current inversion of the current I.sub.SR, as will now be explained
in more detail.
[0083] Before describing this operation in detail, it should be
pointed out that the operation of the comparator 1700 in the
CYCLE[n+2] in detecting the load current transient through the COMP
signal occurs before the end of the conduction time TSW of the
CYCLE[n+2]. Thus, this approach enables the detection of a load
current transient in the half-cycle SC in which the load current
transient occurs, which correspond to the time when the load
current Iload is at a maximum level. This fast detection of the
load current transient through the comparator 1700 enables the
prevention of current inversion in the half-cycle SC in which the
load current transient is detected, namely in the CYCLE[n+2] in the
example of FIGS. 17A-17F.
[0084] From the above description of the operation the comparator
1700 in relation to CYCLE[n], CYCLE[n+1] and CYCLE[n+2], it is seen
that two situations arise in relation to the output signal COMP
generated by the comparator. The first situation is that
illustrated and described above for CYCLE[n] and CYCLE[n+1], namely
where no load current transient occurs from one half-cycle to the
next. In this situation, the comparator 1700 will activate the COMP
signal within the certain percentage of the conduction time TSW,
namely within .beta.TSW. When this occurs, no action is necessary
to prevent current inversion of the current I.sub.SRi, I.sub.SR2
through the synchronous rectification transistors SR1, SR2 and
these transistors can continue being controlled through the
adaptive turn-off control algorithm or any other suitable turn-off
control algorithm for controlling the synchronous rectifier
transistors in the rectifier 5 of FIG. 6.
[0085] The second situation that arises is that illustrated and
described above for CYCLE[n+2], namely where a load current
transient results in a reduction in a high load current Iload, such
as a maximum or full load current to zero load current transient.
In this situation, the comparator 1700 will not activate the COMP
signal within the certain percentage of the conduction time TSW,
namely within TSW. Where the adaptive turn-off control algorithm is
being used to control the transistors SR1, SR2, the detection of a
load current transient as indicated by the COMP signal (i.e., COMP
signal is not activated within time limit .beta.TSW of the
conduction time TSW), the programmable delay PD, which is also
indicated as delay T.sub.DEL_OFF in FIGS. 16B and 16C, is reset to
zero. This zero delay will result in the transistors SR being
turned OFF in response to the earlier zero-crossing of the voltage
signal Vdvs as previously discussed above in relation to FIGS. 11A,
11B, for example. This turn OFF of the transistors may not result
in the ideal conduction time T.sub.diode-ideal of the body diode
Dbr of the transistors SR, but this will avoid current inversion.
Moreover, the adaptive turn-off control algorithm will, after the
programmable delay PD is reset to zero in response to the COMP
signal indicating a load current transient, operate as described
above in relation to FIGS. 11A and 11B to once again adjust the
programmable delay over consecutive half-cycles SC until the actual
conduction time T.sub.diode of the body diode Dbr is approximately
equal to the ideal conduction time T.sub.diode-ideal.
[0086] As mentioned several times in the above description, the
current inversion prevention control algorithm described herein is
not limited to being utilized in conjunction with the adaptive
turn-off control algorithm. Other control methods for controlling
the turning OFF of the synchronous rectification transistors SR in
a rectification circuit, such as rectifier 5 in FIG. 6, may be
utilized in combination with the current inversion prevention
control algorithm. In another embodiment, the detection of a load
current transient as indicated by the COMP signal is used by
control circuitry, such as the control circuit 7 in FIG. 6, to
place the switching converter 100 (FIG. 6) in a "safe state" for
turn OFF of the transistors SR1, SR2 to thereby avoid current
inversion.
[0087] FIG. 18 is a functional block diagram illustrating a
threshold calculation circuit 1800 of the current inversion
prevention control algorithm for generating the threshold TH
according to one embodiment of the present disclosure. In this
embodiment, an average calculation circuit 1801 senses the voltage
signals V.sub.DVS1, V.sub.DVS2 of the transistors SR1, SR2 from the
previous N half-cycles SC of the switching converter 100.
[0088] More specifically, the average calculation circuit 1801
senses the voltage signals V.sub.DVS1, V.sub.DVS2 during the
conduction times TSW of each of the transistors SR1, SR2 over the
previous N half-cycles SC, and calculates an average voltage signal
V.sub.DVS1_AVG, V.sub.DVS2_AVG for each of these sensed voltage
signals. Thus, the average calculation circuit 1801 generates the
average voltage signal V.sub.DVS1_AVG indicating the average of the
voltage signal V.sub.DVS1 over the prior N half-cycles SC, and also
generates the average voltage signal V.sub.DVS2_AVG indicating the
average of the voltage signal V.sub.DVS2 over the prior N
half-cycles SC.
[0089] A peak calculation circuit 1802 receives the average voltage
signals V.sub.DVS1_AVG and V.sub.DVS2_AVG and generates therefrom
the threshold TH, which includes a separate peak voltage reference
threshold for each of the transistors SR1, SR2 as previously
mentioned. Thus, the peak voltage reference threshold TH includes a
first peak voltage reference threshold TH1=(.alpha.V.sub.DVS1_PK)
and a second peak voltage reference threshold
TH2=(.alpha.V.sub.DVS2_PK) for use by the comparator 1700 in
sensing the voltage signals V.sub.DVS1 and V.sub.DVS2. Accordingly,
although not shown in FIG. 17F, the comparator 1700 compares the
voltage signal V.sub.DVS1 to the first peak voltage reference
threshold TH1=(.alpha.V.sub.DVS1_PK) and compares the voltage
signal V.sub.DVS2 to the second peak voltage reference threshold
TH2=(.alpha.V.sub.DVS2_PK) for use by the comparator 1700 in
sensing the voltage signals V.sub.DVS1 and V.sub.DVS2.
[0090] FIG. 19 is a functional block diagram of one embodiment of
the comparator 1700 of FIG. 17. In the embodiment of FIG. 19, the
comparator 1700 includes a first comparator circuit 1900-1 that
receives the first peak voltage reference threshold
TH1=(.alpha.V.sub.DVS1_PK) and the voltage signal V.sub.DVS1 of the
transistor SR1 and generates a first comparison output signal COMP1
based on a comparison of the two received signals. The comparator
1700 includes a second comparator circuit 1900-2 that receives the
second peak voltage reference threshold TH2=(.alpha.V.sub.DVS2_PK)
and the voltage signal V.sub.DVS2 of the transistor SR2 and
generates a second comparison output signal COMP2 based on a
comparison of the two received signals. Thus, the first comparator
circuit 1900-1 generates the COMP1 signal to indicate whether a
load current transient has been detected in relation to switching
of the transistor SR1 and the second comparator circuit 1900-2
generates the COMP2 signal to indicate whether a load current
transient has been detected in relation to switching of the
transistor SR1. The comparator 1700 of FIGS. 17 and 19 and the
threshold calculation circuit 1800 of FIG. 18 are part of the
control circuit 7 of FIG. 6 and may be implemented through suitable
circuitry including hardware, software, firmware, or a combination
thereof, as will be appreciated by those skilled in the art.
[0091] The above-described embodiments of the present disclosure
are described in the context of a resonant switching converter
including synchronous rectification circuitry including first and
second synchronous rectifier transistors that are MOSFETs.
Embodiments of present disclosure are not limited to this
particular type of switching converter, synchronous rectification
circuitry, or type of synchronous rectifier transistor.
[0092] In other embodiments of the present disclosure, the
synchronous rectification circuitry includes at least one
synchronous rectifier transistor. Moreover, the synchronous
rectifier transistors are bipolar transistors or insulated gate
bipolar junction transistors in other embodiments of the present
disclosure.
[0093] The various embodiments described above can be combined to
provide further embodiments. These and other changes can be made to
the embodiments in light of the above-detailed description. In
general, in the following claims, the terms used should not be
construed to limit the claims to the specific embodiments disclosed
in the specification and the claims, but should be construed to
include all possible embodiments along with the full scope of
equivalents to which such claims are entitled. Accordingly, the
claims are not limited by the disclosure.
* * * * *