BACKGROUND OF THE INVENTION
Field of the Invention
The present invention generally relates to bipolar micro-power rail-to-rail
amplifiers, and more specifically to a micro-power output stage that is capable
of delivering a high ratio of load current to no load idle current.
Description of the Related Art
Battery powered amplifiers may spend long periods of time in an "idle"
condition with no input signal, or with intermittent input signals as in the case
of voice signals. Power dissipated in these idle periods is wasted and may shorten
the lifetime of the battery.
The output stage of a rail-to-rail amplifier is designed to deliver
a specified amount of signal power to a load and to consume low power during the
idle periods. A standard measure for an amplifier's power consumption is its no
load idle current. Present bipolar output stages contribute at least 100µA of no
load idle current and require a relatively large number of components to deliver
the desired signal gain, suitably an open loop gain of one million and a closed
loop gain of one. The amount of power dissipated during idle periods is too high
for applications such as battery powered instrumentation and telecommunication
equipment. Furthermore, the number of components and heat dissipation requirements
increase the physical dimensions of the integrated circuit, which increases its
Analog Devices, Inc. produces a "Precision Rail-to-Rail Input &
Output Operational Amplifier" OP284 that has the ability to swing rail-to-rail
at both the input and output. The OP284 has a low supply current of approximately
800µ per amplifier at room temperature. The OP284's output stage uses a constant
current source to supply the gain current needed to achieve sufficient signal gain,
and hence idles at a relatively high current level.
Analog Devices, Inc. also produces a "Micropower Sin-gle-Supply Rail-to-Rail
Input/Output Op Amp" OP291. The OP291 has a low supply current of 300µ per amplifier
and consumes a quiescent current of 600µ. The OP291's output stage uses output
transistor that have 180x relative emitter sizes to achieve the desired gain. This
requires a lot of space on the chip. Furthermore, the output stage requires a
high number of components.
SUMMARY OF THE INVENTION
The present invention seeks to provide a bipolar micro-power rail-to-rail
operational amplifier having a low complexity output stage that provides a high
ratio of output load current to no load idle current.
This is accomplished with a dual differential input stage that converts
a rail-to-rail input voltage into dual differential currents, a transimpedance
stage that converts the differential currents into a single ended drive voltage
at a gain node, and an output stage that amplifies changes in the drive voltage
to produce a rail-to-rail output voltage at an output terminal.
The output stage includes first and second output transistors of
opposite conductivities whose current circuits are connected in series at the output
terminal between high and low voltage supplies. A gain transistor has a base that
is held at a constant reference voltage, an emitter that is connected to the base
of the npn, and a collector that conducts a gain current. A regenerative current
source supplies current to the gain transistor by returning the gain current in
a regenerative feedback loop to its emitter so that the current source idles at
a low current but is capable of supplying much higher currents. A voltage element
responds to the gain current by applying a control voltage that varies approximately
logarithmically at low gain currents and approximately linearly at high gain currents,
to the base of the second output transistor so that its output current varies approximately
proportional to the gain current at low levels and as an approximately exponential
function of said gain current at high levels.
A control transistor having a base, a collector and an emitter that
is connected to the base of said first output transistor and the emitter of said
gain transistor. The control transistor responds to the drive voltage at its base
by modulating the base-emitter voltages of the first output and gain transistors
in opposite directions so that their output currents are unbalanced thereby generating
a load current at the output terminal which adjusts the output voltage. The control
transistor also conducts a control current that stabilizes the feedback loop.
For a better understanding of the invention, and to show how the
same may be carried into effect, reference will now be made, by way of example,
to the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
DETAILED DESCRIPTION OF THE INVENTION
- FIG. 1 is a block diagram of a micro-power rail-to-rail operational amplifier
in accordance with the present invention;
- FIG. 2 is a simplified schematic diagram of an output stage for the operational
amplifier shown in FIG. 1; and
- FIGs. 3a, 3b and 3c are together a schematic diagram of a preferred embodiment
of the operational amplifier shown in FIG. 1.
As shown in FIG. 1, a micro-power rail-to-rail operational amplifier
(opamp) 10 includes a transconductance input stage 12 that converts a rail-to-rail
differential input voltage signal Vin into dual differential currents
Ilow+, Ilow- and Ihigh+,
Ihigh-, a transimpedance stage 14 that converts the differential
currents into a single ended drive voltage Vd at a gain node 16, and
an output stage 18 that amplifies changes in Vd to produce a rail-to-rail
output voltage Vo at an output terminal 20. As shown, amplifier 10
is a single-sided supply amplifier having a high supply voltage Vcc
of 5V and a low supply voltage Vee of ground potential. An external
load 22 is preferably connected between output terminal 20 and Vee.
The input and transimpedance stages suitably provide 60 dB of gain, and output
stage 18 suitably provides another 60 dB of gain. Hence, amplifier 10, operating
open loop, would produce a 1V output voltage Vo in response to a 1µ
differential input voltage Vin.
In practice, Vin is applied to the input stage's non-inverting
terminal 24 and output terminal 20 is connected to the input stage's inverting
terminal 26 to provide negative feedback that forces output voltage Vo
to substantially follow Vin. An error voltage Ve = (Vin-Vo)
of between approximately 0 and 5µV is maintained across the input terminals so
that the output voltage Vo follows Vin. In this closed loop
configuration, amplifier 10 is a broadband unity gain amplifier with low signal
distortion that delivers a controlled amount of signal power to external load 22.
Amplifier 10 is said to be in balance when it delivers zero load
current. Under no load conditions, amplifier 10 can not source any load current,
and hence is always balanced. Under load conditions, amplifier 10 is only balanced
when Vo=Vee. During idle periods, Vo will fall
to near Vee
but will not be perfectly balanced due to practical limitations
in the output stage. In the preferred embodiment of the invention, amplifier 10
supplies a maximum of approximately 5mA of load current IL to load 22
while drawing a no load idle current of only approximately 50µA so that the amplifier
consumes a low amount of power during idle periods.
Input stage 12 preferably includes a pair of differential input circuits
28 and 30, which have identical circuit topologies except that their transistor
conductivities are reversed. Input circuit 28 operates for input voltages Vin
down to Vee, but shuts down when Vin gets to within two diode
voltages of a high supply voltage Vcc. Similarly, input circuit 30 operates
for input voltages Vin up to Vcc but shuts down when Vin
gets to within two diode voltages of Vee. Together input circuits 28
and 30 allow Vin to swing between approximately the low and high supply
voltages Vee and Vcc, respectively, without clipping. Both
input circuits are turned on over a majority of the input signal range and one
of the two are turned on when Vin approaches one of the supply voltages.
Alternately, the input circuits could be controlled so that only one was turned
on at a time.
Input circuits 28 and 30 produce differential currents Ilow+,
Ilow- and Ihigh+, Ihigh-.
Under no load conditions, amplifier 10 does not supply load current at output terminal
20 and is said to be in balance so that these currents are equal. Under loaded
conditions, the differential currents are imbalanced to supply load current IL
at terminal 20 and maintain Vo at substantially Vin. When
driving load 22, input circuits 28 and 30 respond to changes in Vin,
and hence Ve by adjusting differential currents Ilow+
and Ihigh+ so that they move in opposite directions and adjusting
Ilow- and Ihigh- so that they move
in opposite directions to modulate the load current supplied at terminal 20. The
negative feedback maintains the differential currents at approximately their no
Transimpedance stage 14 responds to changes in the differential currents
produced by input stage 12 by reducing drive voltage Vd over a range
of approximately 0 to 5mV relative to a known reference voltage to produce a drive
current at gain node 16. Input circuit 28 and a high side of transimpedance stage
14 are tied through resistors R1 and R2 to Vcc to supply differential
currents Ilow+, Ilow- to input circuit
28 and currents I1, I2 to the transimpedance stage. Similarly,
input circuit 30 and a low side of transimpedance stage 14 are tied through resistors
R3 and R4 to Vee to draw differential currents Ihigh+,
Ihigh- from input circuit 30 and currents I3,
I4 from the transimpedance stage. Under no load conditions, currents
I1 through I4 are substantially the same and drive voltage
Vd remains at the reference voltage so that the drive current at gain
node 16 remains constant. Under loaded conditions, the differential currents are
imbalanced which reduces drive voltage Vd
and the amount of drive current
supplied at gain node 16.
Transimpedance stage 14 is designed so that the voltages across resistors
R1 and R2 are substantially constant. Therefore, changes in Ilow+
and Ilow- induce equal and opposite changes in I1
and I2. Transimpedance stage 14 forces I3 to be substantially
equal to I1 so that changes in Ilow+ and Ihigh+
adjust the voltage across resistor R3 to reflect the change in Vin.
Transimpedance stage 14 is further designed so that the voltage across
resistor R4 tries to follow the voltage across R3. This causes currents I2
and I4 to move in opposite directions in response to a change in Vin
thereby modulating drive voltage Vd relative to the reference voltage
and adjusting the drive current at gain node 16.
Drive voltage Vd is preferably applied to a buffer circuit
32, which shifts Vd towards the low supply Vee by an amount
substantially equal to the nominal voltage across resistor R3. This is done to
set the value of Vd at balance to the particular reference voltage level
and to reduce the amplifier's offset voltage. Circuit 32 also buffers the transimpedance
stage so that it does not have to directly supply the drive current to output stage
Output stage 18 responds to the level shifted and buffered voltage
signal, also referred to as Vd, by supplying load current IL
at output terminal 20 to produce output voltage signal Vo across external
load 22. Output stage 18 uses a regenerative current source to provide an internal
gain current that is used to control signal gain. The regenerative current source
supplies the gain current in a regenerative feedback loop so that the gain current
can vary over a wide range, approximately 0µA to approximately 110µA, in response
to changes in drive voltage Vd. This allows output stage 18 to supply
the large amount of gain current needed to provide a high gain output stage while
idling under no load conditions at approximately 3µA. Under loaded conditions,
the regenerative current source will idle at a somewhat higher level because in
practice output stage 18 can not be driven all the way to ground, and hence a
small current will flow through the load.
By supplying gain current as it is demanded by changes in Vin,
output stage 18 lowers the average power consumed by the amplifier without sacrificing
performance. In the preferred embodiment amplifier 10 consumes a total no load
idle current of approximately 50µA with only approximately 8µA being consumed
by the output stage. The maximum load current is 5mA so that the ratio of load
current to no load idle current is approximately one thousand. Furthermore, the
regenerative current source is self-limiting, which prevents the gain current from
increasing unchecked in response to an extreme variation in Vd and destroying
the amplifier. The regenerative current source requires relatively few components
and a moderate amount of chip space.
FIG. 2 is a schematic diagram of output stage 18 shown in FIG. 1.
Output stage 18 comprises a 10x pnp output transistor Q1 whose emitter 34 is connected
to Vcc and a 4x npn output transistor Q2 whose emitter 36 is connected
to Vee. Their collectors 38 and 40 are connected together at output
terminal 20. The 10x and 4x designations represent the transistors emitter sizes
relative to standard 1x size, and are chosen to provide proper bias conditions
for the amplifier. In general, current flows from the supply Vcc
transistors Q1 and Q2 and into negative supply Vee. Under no load conditions,
the collector currents IQ1 and IQ2
are equal, suitably 4µA,
so that amplifier 10 neither sinks nor supplies a difference current IL
at output terminal 20. Under load conditions, output stage 18 responds to Vin
by imbalancing the collector currents so that together transistors Q1 and Q2 source
load current IL to load 22 so that Vo follows Vin.
Alternately, if load 22 was connected to Vcc
instead of Vee,
transistors Q1 and Q2 would sink current from the load.
Drive voltage Vd is applied to the base 42 of a 1x npn
control transistor Q3 whose collector 44 is connected to Vcc
a control current Ic and whose emitter 46 is connected to the base 48
of output transistor Q2. Transistors Q2 and Q3 set the reference voltage for Vd
at two base-emitter voltages above Vee. Control transistor Q3 responds
to variations in Vd by adjusting the voltage at base 48 of transistor
Q2, which changes its collector current IQ2. This imbalances the output
transistors' collector currents and adjusts load current IL. Because
the change in voltage at base 48 of transistor Q2 is the same as the change in
drive voltage Vd, the gain provided by transistor Q2 alone is relatively
A 1x npn gain transistor Q4 has an emitter 50 that is connected to
the emitter 46 of control transistor Q3 for the purpose of sensing changes in Vd.
A voltage source Vref
clamps the voltage at base 52 of transistor Q4 at
two base-emitter voltages above Vee so that at balance Vd
and Vref are the same and under load conditions Vd is less
than Vref. Therefore, gain transiator Q4 and output transistor Q2 respond
to a change in Vd by adjusting their base-emitter voltages by equal
but opposite amounts. Transistor Q4 responds to changes in its base-emitter voltage
by adjusting a gain current Ig that flows through its collector 54.
A regenerative current source 55 includes a current mirror CM1 that
mirrors gain Ig from collector 54 of gain transistor Q4 to a current
limiter 58, which in turn supplies current Isup to emitter 50 of gain
transistor Q4. Current mirror CM1 includes a diode D1 that is connected across
the base-emitter junction of a 1x pnp transistor Q5. The base/collector 56 of diode
D1 is connected to collector 54, and the collector 57 of transistor Q5 is connected
to current limiter 58. A constant current source IS1 injects current Is1,
suitably 1µA, into the signal path at collector 57 to increase the amount of current
Ilim= (Ig+Is1) that is supplied to current limiter
58. During normal operation, current limiter 58 multiplies the current Ilim
by a constant factor greater than one, suitably 1.5, and supplies it as current
Isup to emitters 50 and 40 of gain and control transistors Q4 and Q3.
If drive voltage Vd is reduced too much, control transistor
Q3 turns off so that all of the regenerated current is supplied to gain transistor
Q4. Because the signal path has a gain greater than one, gain current Ig
if left unchecked would increase without bound and destroy the amplifier. Therefore,
current limiter 58 responds to large gain currents Ig by reducing its
gain and eventually shutting down to prevent the amplifier from destroying itself.
Current limiter 58 is preferably a current peaking circuit that includes
a 1x npn transistor Q6 having an emitter 59 that is tied to Vee, a base
60 that is connected to collector 57 of transistor Q5, and a collector 61. A resistor
R5, suitably 100 ohms, is connected between base 60 and collector 61. A 1.5x npn
transistor Q7 has an emitter 62 that is connected to Vee, a base 63
that is tied to collector 61 of transistor Q6, and a collector 64 that is returned
to emitter 50 of gain transistor Q4. When gain current Ig is small the
voltage across resistor R5 is very small so that transistors Q6 and Q7 act like
a current mirror with a gain of 1.5 to supply Ig to gain transistor
Q4. As Ig increases, the base-emitter voltage of transistor Q6 increases
logarithmically which tends to cause transistor Q7 to increase Isup.
However, the voltage across resistor R5 increases linearly and drives the voltage
at collector 61 of transistor Q6 towards Vee which tends to reduce Isup.
Thus, current limiter 58 begins to limit at the point where the voltage across
resistor R5 increases faster than the base-emitter voltage of transistor Q6. This
reduces the base voltage at transistor Q7, which limits current Ig.
The regenerative feedback loop is described by a pair of equations
that must be satisfied to stabilize output stage 18. The currents summed at emitter
46 of control transistor Q3, ignoring the base current of transistor Q2 and assuming
that the emitter currents of transistors Q3 and Q4 equal their respective collector
currents Ic and Ig, satisfy:
Isup = Ic + Ig
The current Isup supplied by current limiter 58 is described by the
following feedback equation:
Isup = G(Ig+Is1)
where G is the gain of current limiter 58, suitably 1.5 during normal operation
and progressively smaller as limiting occurs.
At balance, current limiter 58 preferably supplies 3µA of current
(Isup=3µA), which is split between transistors Q3 and Q4 so that Ic=2µA
and Ig=1µA. These values satisfy both equations 1 and 2 so that amplifier
10 is stable. Under load conditions, gain transistor Q4 responds to changes in
its base-emitter voltage by adjusting Ig. Regenerative current source
55 senses the change in Ig and adjusts Isup to supply the
required gain current. Control transistor Q3 absorbs any excess current supplied
by the regenerative process by adjusting Ic so that equations 1 and
2 are again satisfied.
A current controlled voltage element 65 is connected between current
mirror CM1 and Vcc so that the current (2Ig) drawn from the
positive supply to supply current mirror CM1 flows through voltage element 65.
Thus, twice the gain current Ig modulates a control voltage Vc
at the base 66 of output transistor Q1. Control voltage Vc moves in
the same direction as the voltage applied to base 48 of output transistor Q2.
Therefore, transistor Q1 adjusts its collector current IQ1 in opposition
to the change in collector current IQ2, which further imbalances output
stage 18 and increases drive current IL.
Voltage element 65 includes a diode D2, which is implemented as a
diode connected 5x pnp transistor, whose base/collector 67 is connected to base
66 of output transistor Q1 and the input of current mirror CM1. A resistor R6,
suitably 2.9K, is connected between Vcc and D2's emitter 68. Thus, the
base-emitter voltage across output transistor Q1 is the sum of the voltages across
resistor R6 and diode D2, which increase linearly and logarithmically, respectively,
as gain current Ig increases.
At balance, the voltages across resistor R6 and diode D2 are approximately
5.8mV and 0.7V, respectively. Therefore, the diode voltage dominates and the resistor
voltage can be ignored. As a result, voltage element 65 mirrors Ig
a fixed amount of gain to output transistor Q1 so that its collector current IQ1
is proportional to gain current Ig.
However, as gain current Ig increases the voltage drop
across resistor R6 becomes progressively more significant, and at some point the
change in control voltage Vc becomes approximately linear. Output transistor
Q1 responds to a linear change in base-emitter voltage by adjusting its collector
current IQ1 as an approximately exponential function of gain current
The effect of this is that output transistor Q1 sources a lot more
current to load 22 for a given change in drive voltage Vd than output
transistor Q2 sinks for an equivalent opposite change in Vd. Thus, output
stage 18 responds asymmetrically to changes in drive voltage Vd and
only provides high gain when sourcing current to the load. Thus, the load is preferably
connected to low supply Vee so that output stage 18 never sinks current.
Alternately, the high gain output transistor could be used to sink current, in
which case the load would be connected to Vcc. In practice, the nonlinear
gain does not matter unless the amplifier is being used for a special application
that requires gain linearity. Furthermore, the benefits of reduced no load idle
current outweigh the minimal effects of gain nonlinearity.
FIGs. 3a, 3b and 3c are together a schematic diagram of a preferred
embodiment of amplifier 10 shown in FIG. 1. The preferred input stage 12 is similar
to those used in operational amplifiers OP-284 and OP-291 produced by Analog Devices,
Inc., the assignee of the present invention. Input circuit 28 comprises a pair
of 2x pnp transistors Q8 and Q9 whose bases 69 and 70 provide the non-inverting
and inverting inputs 24 and 26, respectively. Their collectors 72 and 74 are connected
to Vee. A pair of 2x npn transistors Q10 and Q11 have bases 76 and 78
that are tied together at node 80, and emitters 82 and 84 that are connected to
the respective emitters 86 and 87 of transistors Q8 and Q9. The collectors 88
and 89 of transistors Q10 and Q11 are connected to resistors R1 and R2 and conduct
input currents Ilow+ and Ilow-, respectively.
A pair of diodes D3 and D4, which are implemented as diode connected
1x npn transistors, are connected between node 80 and respective emitters 82 and
84 of transistors Q10 and Q11. Thus, currents Ilow+ and Ilow-
in transistors Q10 and Q11 are twice the currents in D3 and D4, respectively.
A current source IS2 supplies current Is2, suitably 1.5µA, to node 80
to supply transistors Q10 and Q11, and diodes D3 and D4.
At balance, supply current Is2 is split equally between
the two sides of differential input circuit 28 so that Ilow+
and Ilow- conduct 1.5µA. Input circuit 28 responds to a relative
increase in Vin at the base 69 of transistor Q8 (non-inverting terminal
24) by reducing and increasing the voltages across diodes D3 and D4, respectively,
so that a smaller portion of Is2 flows through diode D3 and a larger
portion flows through diode D4. The changes in diode current are reflected into
Ilow+ and Ilow-.
The topology of input circuit 28 allows Vin to swing from
the low supply voltage Vee to within two base-emitter voltages of positive
supply voltage Vcc. The base-collector voltages of transistors Q8 and
Q9 can operate effectively with zero volts. Therefore, Vin can include
Vee without turning input circuit 28 off. Conversely, if Vin
rises to within two base-emitter voltages of Vcc, the transistors begin
The input stage's dual input circuits 28 and 30 are identical except
for the reversed conductivities of their transistors and diodes, and thus a detailed
description of input circuit 30 is omitted. In general, input circuit 30 includes
a pair of npn transistors Q12 and Q13 whose bases are tied to the non-inverting
and inverting terminals, respectively. A current source IS3 draws a reference current
from a pair of diodes D5 and D6. The diodes are connected across the base-emitter
junctions of pnp transistors Q14 and Q15, and respond to changes in Vin
to modulate their collector currents Ihigh+ and Ihigh-
which flow through resistors R3 and R4, respectively.
The preferred transimpedance stage 14 is the folded cascode circuit
used in opamp OP-284 produced by Analog Devices, Inc. The folded cascode circuit
comprises a pair of 2x pnp transistors Q16 and Q17 having bases 90 and 92 that
are connected together at node 94, and emitters 96 and 98 that are connected to
resistors R1 and R2 and conduct currents I1 and I2, respectively.
A current source IS4 draws a bias current Is4, suitably 0.5µA, from
node 94. A pair of 2x npn transistors Q18 and Q19 have bases 100 and 102 that
are connected together at node 104, and emitters 106 and 108 that are connected
to resistors R3 and R4 and conduct currents I3 and I4, respectively.
The collectors 110 and 112 of transistors Q16 and Q18 are connected together and
conduct the same current, I3=I1. The collectors 114 and 116
of transistors Q17 and Q19 are connected at gain node 16 and together source current
at gain node 16 in response to changes in Vin.
The folded cascode circuit establishes a constant voltage drop across
resistors R1 and R2, suitably 50K, by fixing the voltage at node 94. A resistor
R7, suitably 45K, is connected between Vcc and the emitter 118 of a
diode D7, which is implemented as a diode connected 1.5x pnp transistor, whose
base/collector 120 is connected to node 94. Diode D7 and resistor R7 are biased
so that the voltages across resistors R1 and R2 are approximately the same as
the voltage across R7. As a result, 4.5µA of continuous current flows through resistors
R1 and R2.
The folded cascode circuit uses a 1x npn transistor Q20 to control
transistor Q18 and force I3 to be substantially the same as I1.
The base 122 of transistor Q20 is connected to collector 110, its collector 124
is connected to node 94, and its emitter 126 is connected to node 104. If I3
falls below I1, current is driven into base 122 of Q20. This pulls up
base 100 of Q18 so that it increases I3. Conversely, if I3
rises above I1, the base drive to Q20 is reduced. This pulls base 100
of Q18 down, which in turn reduces I3.
The result is that the folded cascode circuit drives current I3
through resistor R3, which together with current Ihigh+ from
input circuit 30 set the voltage across R3. Currents I3 and Ihigh+
respond to changes in Vin by moving in the same direction so that the
voltage across R3 reflects that change. The voltage at base 100 of Q18, and hence
the voltage at base 102 of Q19 move up and down with the voltage across R3 to
keep I3 the same as I1. Transistor Q19 responds to variations
of its base voltage by adjusting I4 so that the voltage across R4 tries
to follow the voltage across R3.
At balance, I1 = I2 = 3µA and Ihigh+
= Ihigh- = 1.5µA. Since the folded cascode circuit forces
I3 to equal I1, I3
is also 3µA. Thus, transistor
Q19 supplies I4 at 3µA so that the voltages across resistors R3 and
R4 are equal. Because currents I2 and I4 are equal, drive
voltage Vd at gain node 16 remains at its reference value and the drive
current provided to the next stage is unchanged.
Under load conditions, amplifier 10 reduces Vd relative
to its reference voltage so that output stage supplies load current to hold Vo
at Vin. Amplifier 10 responds to an increase in Vin by increasing
Ihigh+ and I1, and hence I3, and reducing
I2 and Ihigh-. This increases the voltage across
resistor R3, which in turn increases the voltage at base 102 of transistor Q19.
As the voltage across resistor R3 is increased, the voltage across resistor R4
that is attributable to Ihigh- is reduced. Thus, transistor
Q19 increases I4 to pull up the voltage across resistor R4. This pulls
the voltage at its collector 116 and at gain node 16 down thereby reducing drive
voltage Vd and reducing the drive current. Amplifier 10 responds to
a reduction in Vin
by increasing drive voltage Vd and increasing
the drive current at gain node 16.
Buffer circuit 32 is connected between gain node 16 and base 42 of
control transistor Q3 in output stage 18. The primary purpose of buffer circuit
32 is to establish a voltage drop equal to the voltage across resistor R3 at balance
between collector 116 of transistor Q19 and base 42 of control transistor Q3. This
keeps the voltages at the collectors 112 and 116 of transistors Q18 and Q19 approximately
equal, which lowers the amplifier's offset voltage. A secondary purpose is to
buffer gain node 16 from base 42 of control transistor Q3 so that the change in
base current does not have to be provided entirely by transimpedance stage 14.
Buffer circuit 32 includes a 1x npn transistor Q21 having a base
128 that is connected to gain node 16, a collector 130, and an emitter 132 that
is connected to one side of a resistor R8. A diode D8, implemented as a 1x diode
connected npn transistor, has a base/collector 134 that is connected to base 42
of control transistor Q3 and an emitter 136 that is connected to the other side
of resistor R8. Transistor Q21 and diode D8 are biased so that the voltage drop
across resistor R8 is approximately equal to the voltage across resistor R3 at
A current mirror CM2 mirrors the collector current in transistor
Q21 to the base/collector 134 of diode D8 to provide the base current that drives
control transistor Q3. Therefore, a change in base current at transistor Q3 of
ΔI only requires a change in drive current at gain node 16 of
ΔIβ where β is the current gain of transistor
Q21. Current mirror CM2 includes a diode D9, implemented as a diode connected
1x pnp transistor, and a 1x pnp transistor Q22. The emitters 138 and 140 of diode
D9 and transistor Q22, respectively, are connected to resistor R7 to draw supply
current from Vcc. (This also provides the bias current that sets the
voltage across resistor R7). The base/collector 142 of diode D9 is connected to
the collector 130 of transistor Q21 and to the base 144 of transistor Q22. Transistor
Q22's collector 146 is connected to base/collector 134 of diode D8.
As described above in FIG. 2, output stage 18 responds to drive signal
Vd by supplying load current IL at output terminal 20 to
an external load thereby modulating its output voltage Vo between ground
and 5V. Regenerative current source 55 supplies gain current Ig as needed
to respond to changes in drive voltage Vd and provide the desired gain.
When amplifier 10 is idling, either because input voltage signal Vin
is not being applied or because the signal pauses momentarily, regenerative current
source 55 supplies only a minimal amount of gain current. This lowers the power
consumption of amplifier 10 without sacrificing performance. This saves money,
lengthens the life of amplifier 10, and reduces the physical dimensions of the
Amplifier 10 uses a single bias circuit 148 to establish the reference
voltage level at the base 52 of transistor Q4 in output stage 18, supply current
to buffer circuit 32 and resistor R7, and supply current for diode D7. This is
done to conserve current and lower the amplifier's power consumption during idle
A current source IS5 supplies bias current Is5, suitably
2µA, that flows from high supply Vcc, through 2x diodes D10 and D11
and into Vee. The base/collector 150 of diode D10 is connected to base
52 of transistor Q4. This sets the base voltage at two base-emitter voltages above
the low supply Vee. The base/collector 152 of diode D11 is connected
to the bases 154 and 156 of transistors Q23 and Q24, respectively. Their emitters
158 and 160 are connected to Vee.
Transistor Q23 has an emitter area of 3x, and thus supplies 3µA of
current. This current is split equally between transistor Q21 and diode D8. Transistor
Q24 has. an emitter area of 1.5x, and thus supplies 1.5µA of current that is drawn
from high supply Vcc through resistor R7, diode D7, transistor Q20,
and into its collector 162.
Amplifier 10 conserves current and thus lowers power consumption
by recycling the bias currents for use in multiple parts of the amplifier. However,
sustained extreme inputs could cause amplifier 10 to self-destruct. To prevent
this, a pair of clamps QC1 and QC2 are added to the folded cascode circuit, and
a current limiter QL1 is added to output stage 18.
Clamp QC1 is a 1x npn transistor having a base 164 that is connected
to transistor Q18's collector 112, a collector 166 that is connected to Vcc,
and an emitter 168 that is connected to transistor Q19's collector 116. In normal
operation, the voltages at collectors 112 and 116 are approximately the same so
that QC1 is turned off. However, if transistor Q19's collector voltage drops too
much, QC1 turns on, which pulls Q19 up and prevents it from saturating.
Clamp QC2 is a 1x npn transistor having a base 170 that is connected
to emitter 136 of diode D8, a collector 172 that is tied to Vcc, and
an emitter 174 that is connected to collector 162 of transistor Q24. In normal
operation QC2 is turned off. However, if transistor Q17's voltage pulls up too
much, it may saturate. This pulls up base 170 of clamp QC2, which turns it on.
Thus, the current supplied by transistor Q24 is provided to clamp QC2 instead
of diode D7. This turns the folded cascode circuit off until input voltage Vin
Current limiter QL1 includes a 1x pnp transistor Q25 having a base
176 that is connected to collector 44 of control transistor Q3, an emitter 178
that is tied to Vcc, and a collector 180 that is connected to collector
130 of buffer transistor Q21. A resistor R9 is connected across the base-emitter
junction of transistor Q25. In normal operation, transistor Q25 is off. However,
if drive voltage Vd, and hence the voltage at base 42 of control transistor
Q3 is increased too much, transistor Q25 will turn on and supply the collector
current for transistor Q21. This reduces the current in current mirror CM2 which
in turn reduces the base drive to transistor Q3 and increases its base voltage.
Amplifier 10 uses four capacitors C1-C4 to control its frequency
response. Capacitor C1, suitably 15pF, is connected between output terminal 20
and gain node 16. C1 is the primary compensation capacitor and serves three functions.
First, C1 provides local negative feedback from the output of the amplifier back
to base 42 of control transistor Q3. C1 also provides pole splitting that drives
the low and high frequency poles of the amplifier outside its frequency range.
Finally, C1 controls the amplifier's slew rate. Capacitor C2 is connected from
transistor Q18's collector 112 to Vee. C2 reduces the high frequency
transconductance of the folded cascode circuit and limits the amplifier's bandwidth.
Capacitor C3 is connected between gain node 16 and base/collector 134 of diode
D8. C3 stabilizes the amplifier by compensating for the parasitic capacitance
at the collector of Q23. Capacitor C4 is connected between gain node 16 and emitter
46 of control transistor Q3. C4 prevents the amplifier from oscillating.
While several illustrative embodiments of the invention have been
shown and described, numerous variations and alternate embodiment will occur to
those skilled in the art. For example, different input and transimpedance stages
could be used with the output stage of the present invention. Such variations
and alternate embodiments are contemplated, and can be made without departing from
the scope of the invention as defined in the appended claims.