The present invention relates to a circuit and method for providing
a high-speed switching regulator. More particularly, the invention relates to
circuit and method that provides a high-speed, bipolar integrated circuit switching
regulator that can switch at megahertz frequencies, yet operates at efficiencies
comparable to or better than previously available bipolar switching regulators
operating at much slower speeds.
With battery powered portable computers on the rise, the demand for
efficient DC-to-DC power conversion is increasing. Battery life is of paramount
importance to portable systems. Other important considerations are reducing space,
weight and cost.
Switching regulators have long been known as offering a means for
converting battery voltage in portable systems to other voltages at high efficiencies.
However, a problem with previously available bipolar integrated circuit regulators
has been that the switch operates at relatively low frequencies (e.g., in the neighborhood
of 100 kHz). These low frequencies, while enabling the circuitry to operate with
reasonable efficiency, require the use of relatively large external inductors and
other components to form the switching regulator. Previously available high-speed,
bipolar switching regulators that use smaller external components have suffered
from a problem of inefficiency, and thus consume too much battery power.
US 4, 947,055 discloses a drive circuit for rapidly switching driver
and driven transistors connected in a Darlington configuration. This circuit includes
a first controllable switch coupled to a control electrode of a driver transistor,
an inductor coupled to the first controllable switch and the control electrode
of the driver transistor, and a second controllable switch coupled to the control
electrode of at least one of the transistors.
US 5,138,202 discloses a proportional base drive circuit for a power
In view of the foregoing, it would be desirable to provide an improved
integrated circuit, bipolar switching regulator that operates at high-frequencies,
in order to reduce the space, weight and cost of external inductors and other
components used with the regulator.
It would further be desirable to provide such an integrated circuit
switching regulator that operates at high efficiency, in order to prolong battery
life in portable systems.
In view of the foregoing, it is an object of the present invention
to provide an improved integrated circuit, bipolar switching regulator that operates
at high-frequencies, in order to reduce the space, weight and cost of external
inductors and other components used with the regulator.
It is a further object of the present invention to provide such an
integrated circuit switching regulator that operates at high efficiency, in order
to prolong battery life in portable systems.
These, and other objects of the present invention, are accomolished
by an improved integrated circuit, bipolar switching regulator circuit as disclosed
in the claims. The circuit can operate in the megahertz range, yet at efficiencies
comparable to or better than switching regulators heretofore available that operated
at much lower frequencies. The circuit provides three switch drive currents: a
first (nominal) continuous current, kept low or at zero in order to conserve power;
a second (boosted) current provided while the switch is transitioning from off
to on in order to increase the speed at which the switching element turns on; and
a third (drive) current provided after the switch has turned on to maintain the
switch at a desired point in saturation. Also to conserve power, the drive and
or boost currents vary as a function of the load on the switch. Additional circuitry
momentarily boosts the base discharge current of the switch while the switch is
transitioning from on to off, in order to increase the speed at which the switch
turns off in an efficient, power-conserving manner.
The circuits of the present invention further increases the speed
at which the switch can transition by enabling the drive current prior to turning
on the switch (so that when the switch is turned on, its drive current has already
ramped up). Additional circuitry is provided by the present invention to enable
the regulator to regulate dual polarity outputs (either positive or negative) using
a common error amplifier (to reduce component count and circuit complexity). The
circuitry also provides a multifunction node which can be used to shut down the
regulator or synchronize the regulatory. Further circuitry is provided for improving
the recovery from output overshoot conditions. Finally, an improved clamp that
prevents the switch from operating too far in saturation (which would slow the
switch down and reduce efficiency) increases the stability of the antisaturation
loop as compared with previously known designs.
The above and other objects of the present invention will be apparent
upon consideration of the following detailed description, taken in conjunction
with the accompanying drawings, in which like reference characters refer to like
parts throughout, and in which:
- FIG. 1 is a schematic block diagram of a current mode switching regulator incorporating
principles of the present invention;
- FIG. 2 is a schematic diagram of an exemplary embodiment of an integrated circuit
switching regulator incorporating principles of the present invention;
- FIG. 3 is a simplified schematic diagram of a prior art circuit configuration
for reducing switch drive current during switch off times;
- FIG. 4 is a graph showing switch driver currents as a function of time for
the circuit of FIG. 2;
- FIG. 5 is a schematic diagram of an exemplary embodiment of an oscillator for
generating timing signals for the circuit of FIG. 2;
- FIG. 6 is a schematic diagram of an exemplary logic circuit for generating
control signals used by the circuit of FIG. 2;
- FIG. 7 is a graph showing outputs produced by the oscillator circuit of FIG.
- FIG. 8 is a schematic diagram of an exemplary shutdown circuit for shutting
down the circuitry of FIG. 2;
- FIG. 9 is a schematic diagram of an exemplary error amplifier and negative
feedback network of the circuitry of FIG. 1;
- FIG. 10 is a graph showing relationship between the synchronization function
and the shutdown function of the circuitry of FIGS. 5 and 8;
- FIG. 11 is a graph showing the voltages on the collectors of transistors 254,
256 and 114 in FIG. 2 against time; and
- FIG. 12 is a graph plotting the voltage at the collector of the switch of FIG.
2 against time, and showing the increased stability of the switch clamp circuitry
of the present invention.
FIG. 1 shows a schematic block diagram of a current mode switching
regulator which incorporates the principles of the present invention. Although
the principles of the invention are described with respect to a current mode regulator,
it will be apparent to persons skilled in the art that many of the principles described
below are equally applicable to other regulator circuits.
Switching regulator 100 may include a low dropout regulator 102,
a trimmed oscillator 104, a trimmed reference voltage 106, an error amplifier 108,
a current amplifier 110, a current comparator 112 to drive a switch 114. As is
described below, in accordance with the principles of the present invention, regulator
100 also may include logic circuitry 116 and driver circuitry 118 to drive switch
114, as well as drive boost circuitry 120. Additionally, regulator 100 may include
combined shutdown and synchronization circuitry 122, which utilizes a single input
pin, a negative feedback regulation network 124 and an oscillator frequency shifting
network 126. As described in more detail in FIG. 9, negative feedback network 124
may include a feedback amplifier 128, resistor 130 and 132, and a means of blocking
the output of the amplifier when its output goes low, shown in FIG. 1 as a simple
diode 134. Current amplifier 110 measures the current passing through switch 114
by using a low value resistor 136 (e.g., 0.1 ohms). The function of nodes SD/SYNC,
NFB, FB, VC, SWO and VIN will also be apparent from the detailed
FIG. 2 shows a detailed schematic diagram of an exemplary embodiment
of the driver portion of an integrated circuit switching regulator incorporating
principles of the present invention. The circuit is designed for coupling to a
source of DC input voltage applied across terminals VIN and GND.
In the circuit of FIG. 2, the switch is comprised of power NPN transistor
114. Drive current for the switch is provided by PNP transistor 226E in combination
with transistor 250. Terminal SWO is the switch output terminal for connection
to an external load (not shown) that, in turn, is coupled to a source of electrical
potential. This external load typically would include one or more current-steering
diodes, an inductive element, and other components arranged in conventional fashion
(e.g., in buck, flyback or boost configurations) to implement a complete switching
regulator. Terminal ISW is a node where measurements of the current passing through
switch 114 may be made (by using resistor 136). For example, when the circuit shown
in FIG. 2 is used in a current mode switching regulator, the voltage across resistor
136 is used to determine when to shut off switch 114. The measured voltage is compared
to a reference voltage which is set to represent the current shut-off point. When
the measured voltage exceeds the reference voltage, the current trip point has
been reached and switch 114 is shut-off (when the circuit has been shut down or
a high impedance condition exists, resistor 224 prevents excess leakage current
from inadvertently turning on switch 114).
Switch 114 is turned on and off by a signal, SWON ("SWitch ON"),
via NPN transistors 240, 242, 254 and 256. When SWON is de-asserted (i.e., it is
low), transistors 240 and 254 are off. This allows current provided by the collectors
of PNP transistors 226C and 226D to drive the bases of transistors 242 and 256,
turning those transistors on. (As further discussed below, transistors 226A-E form
a controllable current source. Although shown as five separate transistors, persons
skilled in the art will recognize that transistors 226A-E may be implemented instead,
as in an actual embodiment of the circuit, as a single transistor 226 having multiple
collectors A-E. In either case, transistor 226 is kept off during shutdown or high
impedance conditions by resistor 202 which is connected to its base.) When transistor
242 is on, the base of transistor 250 is pulled low to turn that transistor off
(transistor 250 is kept off by resistor 220). This prevents base drive from reaching
transistor 114, and so the switch is off. As further discussed below, transistor
256 helps to turn and maintain switch 114 off (and diode 264 is used to increase
the turn off speed of transistor 256).
When signal SWON is asserted high, the opposite occurs. Transistors
240 and 254 are turned on, causing the currents provided by the collectors of transistors
226C and 226D to be shunted to ground. The bases of transistors 240 and 254 are
connected to ballast resistors 214 and 216, respectively, to prevent one saturated
transistor from driving the other into saturation. Transistors 242 and 256, accordingly,
are turned off. As a result, collector current from transistor 226E drives the
base of transistor 250 through resistor 218 to turn that transistor on. Transistor
250 and PNP transistor 226E thus drive the base of switch 114 to turn the switch
The amount of current required from transistor 226E depends on whether
switch 114 is conducting a large or small current. When the load on switch 114
is high, transistor 226E should preferably supply a relatively large current sufficient
to cause the switch to turn on and to be driven to a desired operating point. The
precise amount of drive current required to accomplish this will vary depending
on the load on switch 114. Too much drive current is a waste of power and reduces
efficiency. Too little drive current prevents the switch from turning on completely
and would leave the switch in an undesirable high power dissipation state. When
switch 114 is off, on the other hand, transistor 226E preferably should provide
only a nominal or zero current.
In view of the foregoing, and in accordance with an aspect of the
present invention, the circuit of FIG. 2 thus includes circuitry for reducing switch
drive current during periods of time that switch 114 is off. This circuitry includes
transistors 236, 238 and 228 as well as resistors 208, 210, 212 and 204, operating
in conjunction with signal SWDR. The circuitry operates as follows.
When switch 114 is off (i.e., when signal SWON is low), signal SWDR
("SWitch DRive") is high. Thus, no current flows through transistor 236, resistor
210 or transistor 238. To keep the PNP current source circuitry of transistors
226A-E biased, resistor 204 provides a path for a nominal current to flow through
current-setting transistor 226A. NPN transistor 228, driven by reference voltage
VB (which, in the exemplary circuit of FIG. 2, is preferably 2.2 volts),
drives the base of transistor 226A until transistor 226A's collector current is
sufficient to support the load (resistor 204) on transistor 226A's collector. Reference
voltage VB may be generated by circuitry, not shown, in any of a number
of conventional ways that will be readily apparent to those skilled in the art.
For example, a bandgap reference circuit may be connected to an amplifier to adjust
the bandgap reference voltage VBG (typically 1.24 volts) to the desired
reference voltage (e.g., 2.2 volts). The voltage at the collector of transistor
226A thus is equal to the magnitude of VB (2.2 volts) minus the base-emitter
voltage of transistor 228 (about 0.7 volts). The setting by resistor 204 of transistor
226A's nominal current sets the nominal currents provided by transistors 226B-E.
This is because the collector currents of transistors 226B-E are related to the
collector current of transistor 226A by the ratios of the areas of the transistors.
As indicated in the exemplary circuit of FIG. 2, the ratios of the currents A:B:C:D:E
of transistors 226A-E are 1x : .2x : .4x : 1.6x : 5x, respectively.
One known way in which switch drive current can be reduced during
switch off time is shown in FIG. 3. In this configuration, the circuit of FIG.
1 has been modified so that driver transistor 226' is now part of the SWON signal
path via NPN transistor Qx (one circuit with such a configuration is
shown in LT1074/1076 Step-Down Switching Regulator," 1992 Linear Databook Supplement,
Linear Technology Corporation, pp.4-193 to 4-207 (1992)). In this circuit, when
signal SWON is low, switch driver 226' is off. Thus, no drive current is provided
to switch 114. This circuit is less desirable than the implementation of FIG. 2,
however, because bipolar integrated circuit PNP transistors are slow. Particularly
in high speed switching regulator circuits, where the switching element operates
in the megahertz range, the length of time it takes transistor 226' to turn on
would cause the switch to turn on too slowly. This would result in the switch having
poor AC turn-on characteristics, a situation that would waste power in the switch.
One problem that can occur when drive current to a switch is reduced
during the switch's off time, as described above, is that the amount of time needed
to turn the switch back on is increased. This is because it takes time for the
drive current to ramp up, which causes the switch itself to turn on more slowly
than it otherwise would (because the speed at which a transistor switches is related
to the magnitude of its drive). To solve this problem, another aspect of the present
invention provides circuitry that enables the base drive current for the switch
prior to when the switch itself is turned on. Pre-enabling the switch's base drive
allows the drive current to increase to the proper magnitude before being applied
to the switch. Because the speed at which switch 114 turns on is related to the
magnitude of its drive current, pre-enabling the drive current results in the switch
turning on more quickly than it otherwise would.
The drive current for switch 114 is turned on in advance of the switch
by the timing of signals SWDR and SWON. FIG. 4 is a graph depicting the timing
relationship of signals SWDR and SWON, the switching of transistor 114 (shown by
the trace for terminal SWO), and the drive current provided by transistor 226E.
In FIG. 4, signal SWDR goes low at time T0. This enables (turns on)
the switch drive current provided by transistor 226E, which ramps up as shown by
the trace IC(226E). Between times T0 and T1, the
switch drive current is allowed to rise. At time T1, when the drive
current provided by transistor 226 has reached a desired magnitude, signal SWON
goes high to turn on switch 114. Switch 114 turns on at time T2, as
indicated by the falling trace (SWO) of the switch's collector voltage.
FIGS. 5 and 6 show exemplary circuits for generating signals SWON
and SWDR in proper timing relationship to one another. The manner in which these
circuits operate may be understood by reference to the timing chart of FIG. 7.
FIG. 5 depicts a simplified schematic of the regulator's oscillator
circuitry 500, the output of which is a sawtooth (as shown in FIG. 7) having a
rise time that is much slower than its fall time. The trace labelled "CAP" in FIG.
7 shows the oscillator's sawtooth waveform at the output node labelled "CAP" (FIG.
5), while trace "SET" in FIG. 7 is the oscillator's waveform at the node in FIG.
5 labelled "SET." An oscillator cycle in FIG. 7 begins at time T0, when
the oscillator's output begins to fall. It is easier to understand how the oscillator
works, however, by starting the analysis at time T1 when the oscillator's
output begins to rise. NPN transistors 510 and 514 form a current comparator, which
compares the collector current of transistor 510 (current from first current source
502 (I1) passing through resistor 524) to a second reference current
source 504 (I2). At time T1, transistor 510's collector current
is below I2, causing transistor 510 to pull the SET node low. This turns
off transistors 516 and 522. With transistor 522 off, capacitor 526 is charged
by current source 508 (I4). This causes the capacitor's voltage to linearly
increase, as shown in FIG. 7 by the trace labelled CAP between times T1
and T,. This voltage at the CAP node is transferred to the emitter of transistor
520 by transistor 518 (which is kept biased by current source 506 (I3)).
Together, transistors 518 and 520 form a unity-gain buffer. As the voltage on transistor
520's emitter rises, the current through resistor 524 and transistor 510 rises.
The CAP voltage continues to rise until time T3, when transistor 510's
collector current exceeds that of reference current I2 and the voltage
on transistor 510's collector is sufficient to turn on transistor 516. When this
occurs, the voltage at the SET terminal goes high. This turns on transistors 516
and 522, and marks the start of the oscillator's fall time.
When transistor 516 turns on at time T3, it disables the
current comparator formed by transistors 510 and 514 by shunting to ground reference
current I2. At the same time, transistor 522 discharges capacitor 526,
causing the voltages across capacitor 526 and at the emitter of transistor 520
to fall rapidly (see the CAP trace in FIG. 7, between times T3 and T4).
This continues until time T4 is reached, where the voltage on transistor
520's emitter is no longer sufficient to keep transistors 516 and 522 turned on.
When this occurs, transistors 516 and 522 turn off. Transistor 516 thus releases
the current comparator, and the SET terminal goes low again to begin another cycle.
The SET signal produced by oscillator 500 is used by the logic circuitry
of FIG. 6 to generate signals SWON and SWDR. Switch drive current is enabled at
the beginning of oscillator fall time (i.e., at time mark T0 in FIG.
7; see also FIG. 4). The "dead zone" between times T0 and T1
in FIG. 7, when the oscillator's sawtooth is falling, provides a fixed delay to
the point in time (at T,) when the switch is turned on (see FIG. 4). In FIG. 6,
transistors 602 and 604 (which get bias current from current sources 610 (I1)
and 612 (I2), respectively) form a set/reset flip-flop. Prior to time
T0, the flip-flop is reset by assertion of a RESET signal. (The RESET
signal is generated when the switch current exceeds the current trip point described
above for a current mode regulator.) When the RESET signal is asserted momentarily
low, transistor 606 is turned off to cause signal SWDR to go high, transistor 602
is turned off and the SWON signal is pulled low (through diode 616). This causes
the flip-flop to switch to its reset state (with transistor 602 off and transistor
604 on). At time T0, the SET signal from oscillator 500 is asserted
high (see FIG. 7). This turns on transistor 608, which forces its collector low
to turn off transistor 604 via diode 620. The collector of transistor 604 resultantly
goes high, and the flip-flop thus is "set" (transistor 602 is on and transistor
604 is off). This causes transistor 606 to turn on (because current source 612
drives transistor 606's base), and so the collector of transistor 606 goes low
-- which asserts the SWDR signal low (as earlier described). Transistor 608 also
holds signal SWON low via diode 618, so that signal is not asserted at this time.
At time T1, however, the SET signal goes low (see FIG. 7). This turns
off transistor 608, causing its collector to go high to assert signal SWON (which
gets its bias current from current source 614 (I3)). Thus, the circuitry
of FIGS. 5 and 6 turns on the drive current for the switch before the switch itself
is turned on.
The foregoing pre-enabling of the switch drive current enhances the
speed at which the switch transitions from off to on because, as mentioned, the
speed at which a transistor switches on is related to its drive current. As the
drive current increases, the speed of switching increases. However, if the drive
current is too large efficiency will decrease. To increase the speed of switching
of transistor 114 without unnecessarily impacting on the efficiency of the circuit,
and in accordance with still another aspect of the present invention, the circuit
of FIG. 2 includes circuitry to cause the drive current to the switch to be varied
as a function of the amount of drive the switch needs. In addition, the circuitry
momentarily boosts the drive current provided to the switch while the switch is
turning on in order to increase the speed at which this switching occurs. After
the switch has turned on, and while the switch remains on, its drive current is
reduced to a level that varies as a function of the load on the switch. This circuitry
operates as follows.
Referring again to FIG. 2, and as earlier discussed, signal SWDR
is de-asserted (i.e., it is high) while switch 114 is off. Under this condition,
transistor 226E produces a nominal (or zero) first current set by the value of
resistor 204. When signal SWDR signal goes low to turn on the switch drive current,
however, the load placed on the collector of transistor 226A increases to become
the sum of the currents flowing through resistors 204 and 210, plus the collector
currents of transistors 236 and 238. This causes transistor 226E to produce a second
(boosted) current calculated to turn the switch on very rapidly.
Transistor 236 turns on, when signal SWDR goes low, because its base
is driven by voltage VC2. This voltage is an internally buffered version
of the voltage VC which represents the desired switch current trip point.
As previously described, this voltage varies as the load on switch 114 varies.
Because the voltage of the current trip point (VC) is closely related
to the current through switch 114, transistor 236's collector current thus will
follow the switch current. Therefore, the amount of drive switch 114 receives from
transistor 226E will depend on the switch's load current. This results in delivering
to the switch only as much drive current as the switch requires, and conserves
Also when SWDR is asserted transistor 238 is turned on. However,
unlike transistor 236 (which stays on as long as switch 114 is on), transistor
238 stays on substantially only during the period of time that switch 114 is transitioning
from off to on. Transistor 238 thus causes transistor 226E to provide a boosted
drive current to switch 114 for a momentary duration only when it is needed --
namely, when the switch is turning on in order to increase its switching speed.
Transistor 238 turns on momentarily in this fashion when signal SWDR goes low because
the base of transistor 238 is driven by transistor 232 and a small bias current
from transistor 226B, which clamps transistor 238's base to the reference voltage
VB. (A unity gain buffer formed by transistors 230 and 232 isolates
VB from the base of transistor 238. The bias for transistor 230 is set
by resistor 206.) Thus, when SWDR goes low, transistor 238 turns on. While transistor
114 is off, transistor 244 is also off (because terminal SWO, when coupled to a
load, is high). However, when switch 114 turns on its collector voltage drops.
This turns transistor 244 on, which pulls down on the base of transistor 238 to
turn it off. With transistor 238 off, the load on transistor 226A is reduced and
the boosting of the switch's drive current ceases. The current provided by transistor
226E is thus reduced to a level determined by the current flowing through resistor
204, transistor 236 and resistor 210.
The driver circuitry of FIG. 2, described above, accordingly provides
three different and independent currents: a first (nominal) current provided while
the switch is off in order to conserve power; a second (drive) current provided
while the switch is on to drive the switch to a desired operating point, as a function
of the switch's load current; and a third (boosted drive) current provided substantially
only while the switch is in the process of turning on in order to increase the
speed with which that transition occurs. By virtue of this tri-state drive circuitry,
switching speeds are increased with an actual reduction in the amount of power
Additional circuitry shown in FIGS. 5 and 8 is directed toward another
aspect of the present invention in which a single input node performs a combined
synchronization and shutdown function. FIG. 5 shows synchronization circuit 528
being coupled to oscillator 500. The SYNC/SD node may be an external pin that,
when a synchronization signal is applied, forces the regulator to operate at a
frequency different than the regulator's natural frequency. External synchronization
may be used, for example, to synchronize multiple regulators operating in a single
When the SYNC/SD node is tied high or floats, the regulator circuit
operates at its natural frequency because current source 540 (I6) drives
the base of transistor 532, keeping it turned on. While transistor 532 is on, it
shunts current source 538 (I5) so that transistor 530 remains off. Keeping
transistor 530 turned off effectively isolates synchronization circuit 528 from
the oscillator circuit. When a synchronization signal above the internal oscillator
frequency is applied to the SYNC/SD node, it forces the oscillator to operate at
the higher frequency. This occurs due to the charging and discharging of capacitor
536. The falling edge of the synchronization signal level shifts capacitor 536
so that the voltage at the base of transistor 532 drops. This causes transistor
532 to turn off, which in turn causes current source 538 to turn on transistor
530. When transistor 532 turns on, the reference signal from current source 504
(I2) is shunted (through transistor 530) so that the oscillator is restarted
(i.e., when the collector current on transistor 510 exceeds current I2,
the SET terminal goes high, turning on transistors 516 and 522, thus marking the
start of the oscillator's fall time). While transistor 532 is off, current source
540 charges capacitor 536 until the voltage at the base of transistor 532 is high
enough so that it turns back on, forming a re-triggerable one shot (and the cycle
begins again). Additionally, to improve the synchronization characteristics of
circuit 528, a diode connected transistor 534 is included for clamping the operational
range of capacitor 536. In this manner, capacitor 536 is typically restricted to
operate between +VBE
The SYNC/SD node going low causes multiple things to happen. In the
circuitry of FIG. 5 (as previously described), the oscillator is synchronized.
Additionally, this causes a shutdown signal to be applied to shutdown circuitry
800, as shown in FIG. 8. Current source 818 (I3) provides an internal
shutdown delay (see FIG. 10) which prevents instantaneous shutdown, while also
permitting the synchronization circuitry to also use the SYNC/SD node. If transistor
808 is on, it shunts current source 818 and prevents capacitor 812 from charging.
Transistors 804 and 806 form a current mirror such that their collectors attempt
to draw current in a 5:1 ratio (note that the emitter area ratios are 5:1, respectively).
Current source 816 (I2) is set to supply .3 microamps through transistor
806, so 1.5 microamps should go through the emitter of transistor 804 (due to the
described area ratios). Current source 814 (I1) provides 1.2 microamps
to transistors 802 and 804, as long as the SYNC/SD node is not tied to ground.
Therefore, transistors 802, 804 and 806 are all on and approximately 1.8 microamps
passes through the node formed by the emitters of transistors 804 and 806 (transistor
802 level shifts the threshold at which transistor 808 turns on and off depending
on the status of the SYNC/SD node). This current is more than the current pulled
by current source 820 (I4) (for example, 0.6 microamps) so that transistor
808 is turned on.
A synchronization signal applied to the SYNC/SD node (as shown by
the top trace on FIG. 10) causes the node to go to ground periodically, which shunts
current source 814. This turns off transistors 802 and 804 so that only 0.3 microamps
flows from the emitter node (i.e., the 0.3 microamps from source 816 through transistor
806). Because this current is not enough to overcome current source 820, transistor
808 turns off and capacitor 812 begins charging (see the bottom trace in FIG. 10,
from time T0 to time T1). Once the synchronization signal rises, capacitor 812
is discharged and the cycle begins again. When the SYNC/SD node is tied to ground,
transistor 808 remains off which permits capacitor 812 to charge up (see the bottom
trace in FIG. 10 from time T3 to time T4) until transistor 810 is turned on and
shutdown is accomplished.
Additional circuitry shown in FIG. 8 generates the bandgap reference
voltage (VBG) which is utilized by many of the circuits described herein.
Therefore, all that needs to happen to shutdown the regulator circuit (placing
it in a high impedance state to significantly reduce power consumption) is to shut
down the bandgap generator. The bandgap generator is essentially formed by transistors
822, 824, 826 and 828 (transistors 822 and 824 form a current mirror). Transistors
826 and 828 are provided with an emitter area ratio of 10:1 so that a 60 millivolt
differential is created between the emitters. This voltage is placed across resistor
838 to establish a current into resistor 844, which is selected to get the desired
bandgap voltage (e.g., 1.24 volts). Transistor 830C drives the bandgap circuit.
The current through the collector of transistor 830B (a voltage approximately equal
to VBG-transistor 832's VBE is generated across resistor 840
to create the current through transistor 830B) is mirrored through transistor
830C's collector, as long as the bandgap circuit is on (excess drive current from
transistor 830C is shunted by transistor 842). When capacitor 812 is permitted
to ramp all the way up (i.e., SYNC/SD stays low), transistor 810 is turned on and
current is pulled through buffer transistor 834 so that the bandgap goes low.
When the bandgap goes low, the emitter of transistor 832 goes low and current no
longer flows through transistor 830B. Thus, the regulator circuit may be shutdown
from the same node as the synchronization signal is applied.
The present invention also includes circuitry which enables the regulator
to regulate either a positive or negative voltage. In applications where dual polarity
regulation is required, prior art circuits required several additional pins to
accomplish all the desired functions and also required additional external components.
The circuitry of the present invention utilizes only one additional pin and needs
no external components. FIG. 9 shows a detailed schematic of error amplifier 108,
including overshoot recovery circuit 900, and negative feedback network 124. Error
amplifier 108 is formed by the differential pair of transistors 902 (the negative
input) and 904 (the positive input, which is tied to VBG). The error
amplifier receives its current from a current source formed by resistor 920 and
transistors 914, 916, 918C and 918D (resistor 920 and transistor 918D set the currents
for the circuit shown in FIG. 9). Transistors 902 and 904 have their collectors
coupled to transistors 906A and 908A (diode connected transistors) which have their
bases coupled to transistors 906C and 908C. Transistor 906C is collector coupled
to transistors 910 and 912 which form a current mirror, where the output of the
amplifier is taken from the node between the collectors of transistors 908C and
912 (VC). In this manner, any imbalance between FB and VBG
is mirrored from transistors 902 and 904 through transistors 906A and 908A to the
current mirror, so that the current in the collectors of transistors 908C and 912
will not be equal. The difference is fed via VC to comparator 112 which
adjusts the current shut off point.
The negative feedback network includes differential amplifier transistors
922 and 924, current mirror transistors 926 and 928, current source transistor
930, level shift transistors 932 and 934, input transistors 936 (the positive input,
tied to ground) and 938 (the negative input, tied to the node between resistors
130 and 132), and output transistor 940. As in error amplifier 108 discussed above,
if the inputs to the amplifier are balanced, the currents from the collectors of
transistors 922 and 926 will cancel each other so that there is little net current
on the base of output transistor 940 (sufficient only to drive node FB to VBG).
One of the advantages of the negative feedback network of the present
invention is the fact that the regulator uses the same error amplifier to regulate
positive and negative input voltages. The network circuitry is configured such
that the negative feedback amplifier removes itself from the circuit when positive
voltages are being regulated, without affecting the regulation. When positive voltages
are regulated, the FB node is connected to a positive regulator network, which
regulates FB to a reference voltage (VREF), while the NFB node floats.
VB (which is now equal to VREF) is also provided to the negative
input of the negative feedback amplifier, through resistor 132, causing the output
of the negative amplifier to turn off and isolating the negative amplifier from
the error amplifier. (The negative input has a positive voltage and the positive
input is at ground, therefore the output tries to pull node FB down, but cannot
do so because transistor 940 is a PNP (acting like diode 134 of FIG. 1). During
negative voltage regulation, the FB node floats and the NFB node is connected to
a negative regulator network which regulates NFB at -2VREF. In this
case, the negative feedback amplifier drives the FB node to about VREF
(because transistor 940 is now turned on). Thus, positive and negative voltage
regulation is accomplished using a common error amplifier which retains its multiple
functions including oscillator frequency shifting, overshoot improvement and loop
Another aspect of the present invention shown in FIG. 9 is overshoot
recovery circuit 900 which limits overshoot that regulators typically experience,
either during startup or following the release of an overload condition. Overshoot
recovery circuit 900 includes transistors 906B and 908B (which have a common base
with transistors 906A/906C and 908A/908C, respectively, current mirror transistors
942 and 944, control transistor 946, emitter degeneration resistors 948 and 950
and shunting resistor 952. The circuit operates as follows.
During normal conditions, the currents in the differential pair of
error amplifier 108 (transistors 902 and 904) are mirrored to transistors 942 and
944 (as the currents were mirrored through transistors 906 and 908 to transistors
910 and 912 in the error amplifier). In a balanced error amplifier condition,
transistors 942 and 944 are driven with equal currents, but because of their unequal
emitter areas (the ratio between transistors 942 and 944 may be about 2:3) transistor
944 is pushed into saturation. This causes the collector of transistor 944 to be
low and, therefore, transistor 946 is turned off. While transistor 946 is turned
off, overshoot recovery circuit 900 is essentially disconnected from the error
amplifier. When an overshoot condition occurs, either at startup or any other time,
the current differential appearing on differential pair 902/904 causes transistor
944 to come out of saturation. Transistor 944's collector goes high and turns on
transistor 946 which causes extra current to be pulled from the VC node (resistor
952 controls the amount of current that transistor 946 pulls from VC). The trip
point at which the overshoot recovery circuit becomes active is set by resistors
948 and 950 and the emitter area ratios of transistors 942 and 944. One advantage
of this technique over conventional techniques is that the point at which high
negative slew currents on the VC node are permitted is moved away from the regulating
point. For example, the trip point may be set to VREF plus 50 millivolts.
Another aspect of the present invention, for efficiently increasing
the speed of operation of switch 114, is a circuit that decreases switch turn off
time. In FIG. 2, this circuit includes a network comprising capacitor 258, resistor
222, diode-connected transistor 252, and diode 260. The circuit works as follows.
The speed with which transistor 114 turns off is related to its base
discharge current. Generally, as the base discharge current of a transistor increases
the speed at which the transistor turns off increases. The circuit including capacitor
258, resistor 222 and diode 260 takes advantage of this principle by momentarily
boosting switch 114's base discharge current while the switch is transitioning
from on to off, in order to increase the speed with which that transition occurs.
Once the transistor has turned off, the boosting ceases. This is shown in FIG.
11, which plots against time the voltages on the collectors of transistors 254,
256 and 114 in relation to the base (IB) and collector (IC)
currents of transistor 256 and the current through capacitor 258.
In FIG. 11, the process of turning switch 114 off commences when
signal SWON transitions from high to low. When this happens, as the figure shows,
transistor 254 and transistor 256 will turn on to provide a base discharge current
for switch 114. The current that transistor 256's collector can remove from transistor
114's base is limited, however, by the amount of current that transistor 226D delivers
to transistor 256's base. Capacitor 258, resistor 222 and diode 258 operate to
augment this base current when switch 114's collector voltage begins to rise (which
occurs when the switch begins its transition from on to off, as shown by the trace
labelled VCQ114 in FIG. 11). This additional current through capacitor
258 (see FIG. 11, trace I258) increases transistor 256's collector current
(FIG. 11, trace ICQ256/25), to increase the base discharge current of
switch 114. As transistor 114's collector continues rising, capacitor 258, resistor
222 and diode 260 continue to boost transistor 256's collector current. The result
is that the time it takes switch 114 to turn off is reduced.
The boosted discharge current, fed through capacitor 258, is only
required on an AC basis. It is also required only when switch 114 is being turned
off (i.e., when switch 114's collector voltage is rising). Diode 252 is used,
accordingly, to block current when switch 114 is being turned on. Resistor 222
limits the amount of the current boost. Diode-connected transistor 252, shown coupled
between ground and the anode of diode 260, is provided to discharge capacitor 258
on each ON cycle of switch 114 to prevent the capacitor from peak detecting the
collector voltage of the switch.
By boosting transistor 114's base discharge current only when required
(i.e., while switch 114 is transitioning from on to off), the circuit of the present
invention further increases switching speed without unnecessarily wasting power.
Persons skilled in the art will recognize that the discharge current for transistor
114 could have been increased by increasing transistor 226D's collector current.
However, permanently increasing this current wastes power because the boosted current
is only needed for a short time.
Still another aspect of the present invention that increases the speed
at which transistor 114 switches off and improves efficiency is an improved clamp
for defining the switch voltage in the "on" state. It is well known that the speed
at which a transistor can turn off while operating near or in saturation is related
to how far in saturation the transistor is operating. A transistor operating out
of saturation, or in quasi-saturation, will turn off more quickly than will a more
saturated one. For this reason, it is well known to use a Baker clamp to prevent
a switch from becoming too saturated. Examples of previously used Baker clamp
circuits may be found in United States patent 4,755,741 (see FIG. 6); "Collector
Diffusion Isolation - A new Bipolar Process for Integrated Circuits," Ferranti
Ltd, 2nd Edition, October 1972; and Bob Mammano, "Simplifying Converter Design
with a new Integrated Regulating Pulse Width Modulator," Vol. III of the Proceedings
of Powercon 3: Third National Solid-State Power Conversion Conference, June 1976.
However, traditional Baker clamps -- because their operation relies
on feedback -- can cause instabilities in the circuit in the nature of ringing
or oscillations. This ringing causes electromagnetic interference and may reduce
efficiency of the regulator circuit.
An aspect of the present invention solves this problem in prior art
Baker clamped switches by the circuitry shown in FIG. 2. The Baker clamp in FIG.
2 is comprised of PNP transistor 248 and diode-connected transistor 246. The base-emitter
circuits of these components, when coupled as shown, form a loop with the base
emitter circuit of transistor 250 and the base-collector circuit of transistor
114. The base-emitter voltages of transistors 246 and 248 are chosen (by area-ratioing)
to limit the collector-emitter voltage of switch 114 in the on state. Transistors
246 and 248 become forward-biased -- causing current to be shunted away from the
base of transistor 250. Transistors 246 and 248 thus operate to limit the on state
voltage of switch 114. When the collector-emitter voltage of switch 114 drops below
a desired level, the feedback action of transistors 246 and 248 will cause the
collector voltage of transistor 114 to drop just low enough to turn on both transistors
246 and 248. Hence, transistor 114's collector voltage is clamped, and the operating
point of the switch is limited. This limiting function may be used, for example,
to control the depth of saturation of the switch.
A problem with the circuit of FIG. 2 to extent described so far is
that, as mentioned, the Baker clamp is susceptible to instabilities. These instabilities
are caused by delays in the feedback loop, a major cause of which is a significant
delay associated with diode-connected transistor 246. The present invention solves
this ringing problem by adding a capacitor, CF, across diode-connected
transistor 246 as shown in FIG. 2. The degree to which capacitor CF
ringing depends on the value chosen for the capacitor. FIG. 12 shows results that
might be expected for different values of capacitor CF in the circuit
of FIG. 1, but the effect of the capacitor in other circuits will vary based upon
the selected operating region for the switch. For example, a slight improvement
over the use of no capacitor at all results with a capacitance value of 5 pf (shown
in FIG. 12 by the trace labelled "5 pf"). A value for capacitor CF of
10 pf provides a further improvement in stability as shown by the trace labelled
"10 pf." On the other hand, a value of 20 pf for capacitor CF over-compensates
the circuit, as shown in FIG. 12 by the trace labelled "20 pf." This over-compensation
is undesirable, because it allows the switch to spend more time in a higher power
state. The end result is that the switch, under certain conditions, switches more
slowly, creating AC switching losses. In an actual embodiment of the circuit of
FIG. 2, a value of about 15pf for capacitor CF was found to give satisfactory
The circuits of FIGS. 1, 2, 3, 5, 6, 8 and 9 can be implemented using
commercially available components. For example, the circuits can be constructed
and operated using the components and values set forth in Table 1, below (for the
transistors, only transistor type and area ratio are given):
Resistors 210 and 844:
Resistors 212, 216, 222 and 224:
Resistors 948 and 950:
Transistors 226A, 232, 244, 822, 824, 842, 918D, 926, 928, 936,
938 and 940:
Transistors 226E and 520:
Transistors 228, 240, 510, 514, 914 and 942:
Transistors 230, 516, 530, 532, 534, 602, 802, 806, 808, 810,
828, 832, 902, 904, 916, 922, 924, 930, 932, 934 and 946:
Transistors 236, 238, 252, 518, 604, 606, 804, 836, 910 and
Transistors 242, 254 and 826:
Transistors 522 and 608:
Transistors 834, 906C and 908C:
Transistors 906A, 906B, 908A, 908B and 918C:
Transistors 918A and 918B:
While preferred embodiments of the invention have been set forth
for purposes of the disclosure, modification of these embodiments may occur to
those skilled in the art. For example, while the circuits of the present invention
have been disclosed in the context of a control circuit for a switching regulator,
it will of course be understood by those skilled in the art that the invention
may be employed in any type of circuit utilizing switching transistors. Persons
skilled in the art will appreciate that the present invention can be practiced
by other than the described embodiments, which are presented for purposes of illustration
and. not of limitation.