a

Amplifier Fundamentals
Technical Tutorial – Amplifiers
Basic Level
June 2002

The World Leader in Signal Processing Solutions -1-

 Course Objectives and
Prerequisites…

Course Objective:
To understand the basic AC and DC performance
specifications of operational amplifiers, in order
to be able to recommend and specify an
appropriate amplifier for a given customer
application.
Prerequisite:
There is no pre-requisite to this course

a -2-
What Is an Operational Amplifier?

+
VIN A
- VOUT

„ An Operational Amplifier is an active linear device with a

differential input, a single-ended output and the transfer
function:
VOUT = A * Vin + K
Where A = amplifier “open loop” gain and K = a constant factor.
„ The gain A is typically very high (>>10,000) and not well
defined. The amplifier is normally used with external negative
feedback to define an accurate “closed loop” gain.

a -3-
 Why Do We Need Amplifiers?

“Real” World Signals – Analog Domain Digital Domain

Micro/DSP Processor
Multiplexer

Analog Analog to
Op Amp Signal Digital
Processing Converter

„ An Operational Amplifier is used to amplify and/or level translate

(DC shift) low level, “real world” signals from sensors,
transducers, etc to a high level signal compatible with the input
of the A-D converter.
„ Typical A-D input ranges include 2.5V, 5V and 10V full scale,
either unipolar or bipolar.
a -4-
 “Classic” Amplifier Designs
ROUT

vIN
∆v OUT
vOUT VIN RIN A VIN VOUT
∆v IN

VOLTAGE AMPLIFIER

ROUT
∆V OUT
IIN vOUT
∆IIN
IIN RIN RM IIN VOUT

TRANSIMPEDANCE AMPLIFIER

a -5-
 “Classic” Amplifier Designs (con’t)

ROUT

IIN ∆I OUT
IOUT IIN RIN AIIN IOUT

∆I IN

CURRENT AMPLIFIER

ROUT

∆IOUT
VIN RIN GM VIN IOUT
vIN IOUT
∆V IN

TRANSCONDUCTANCE AMPLIFIER
a -6-
 “Classic” Amplifier Designs (con’t)

VIN VOUT IIN VOUT

Voltage Feedback Transimpedance

„ Both Voltage Feedback and Transimpedance amplifiers are

available in integrated circuit (IC) form.
„ Current and Transconductance amplifiers are generally built with
discrete components

a -7-
 Basic Configurations

Non-Inverting Inverting

+ +

- -
RF

RI RF
RI Gain = 1 + RF/RI
Gain = - RF/RI

a -8-
 Basic Configurations
Z1 Z3

+ Z5 Z7

Z2 Z4
-
Z6 Z8

Unity Gain Buffer (Impedance Isolation)

S1
+
S2 ANALOG-DIGITAL
CONVERTER
S3
-
S4

Switch/MUX to ADC Buffer

a -9-
Additionally, An Op Amp May Be
Used To…

„ Sum 2 or more signals

„ Take the difference of 2 signals

„ Convert a current to voltage

„ Provide the integral of a signal

„ Provide the derivative of a signal

a -10-
 An “ Ideal” Op Amp Has, -

+
ZIN = ∞ ZOUT = zero
-

„ Infinite input impedance

„ Zero output impedance
„ Infinite open loop gain at all frequencies
„ No DC errors
„ Infinite common-mode rejection

a -11-
 A “ Real” Op-Amp…

IB+ EOS
IN+
ZIN ZOUT

IN- IB-

-
EN

A “real” op amp contains a set of finite parameters that are

also sometimes described as “error terms”

a -12-
 Some Key Op Amp Parameters…

„ Typical DC parameters include, -

‹ Input offset voltage, EOS or VOS
‹ Input bias current, Ib

„ Typical AC parameters include, -

‹ Input voltage noise, en
‹ Input current noise, in

‹ Small signal/full power bandwidth

‹ Slew rate

The values of these parameters will determine the applications for which
an amplifier is best suited.

a -13-
 OP Amp Categories

Analog Devices groups its operational amplifiers into 5 basic categories

in order to simplify the selection process:

„ Precision
„ Low bias current
„ Low noise
„ High speed
„ Single supply

a -14-
 Precision Amplifiers

Precision amplifiers are characterized by...

‹ Low initial offset voltages < 100 µV
‹ Low offset drift < 1 µV per deg C

a -15-
 Input Offset Voltage, Eos

EOS
+

„ Modeled as a voltage source in series with the (+) input

terminal.
„ Output voltage error due to offset is Eos x (1 +Rf/Ri)
„ Offset drift vs temperature : µV/deg C
„ Offset drift vs time : µV/month

a -16-
 How Is Offset Voltage Produced?

IC1 IC2

+ IN - IN
+ +
VBE1 - - VBE2

„ Mismatch between VBE of input transistor pair due to

differences in collector currents and transistor geometries
„ Laser-trimmed in factory
„ Bipolar transistors can yield very low offset (< 10 µV) and
drift (<0.1 µV/deg C)

a -17-
 Typical Range of Input Offset
Voltages

Chopper-Stabilized Op Amps < 1 µV

Precision Bipolar Op Amps 10 - 25 µV

Gen’l Purpose Bipolar Op Amps 50 – 500 µV

FET Op Amps 50 – 1,000 µV

High Speed Bipolar Op Amps 100 – 2,000 µV

a -18-
Offset Trim Methods – Factory Trim

(a)

IC1 IC2

IC1 IC2 + IN
+ +
- IN

VBE1 - - VBE2
+ IN - IN
+ + (b)
VBE1 - - VBE2

Laser trimming in the Factory varies the collector currents until VBE1 ≈ VBE2
either by (a) trimming one of the collector currents or by (b) trimming one of
the emitter resistors

a -19-
 Offset Trim Methods – Offset Null
Pins

Trim (“null”) pins are provided on most (but not all) op amps to null out the
offset voltage
a -20-
 Offset Trim Methods – External Trim
Schemes
RB
+VS
ROS
+ +

-VS
+VS - -
ROS
RF

RI RF
-VS RI

RF
Gain = 1 + ------- ; ROS >> RI Gain = - RF/RI ; RB = RF//RI
RI // ROS

a -21-
Input Offset Drift vs Temperature
and Time

„ Vs temperature:
‹ 3 point measurement @ TMIN, 25 deg C and TMAX

‹ Average slope (tempco) : 30 nV/deg C to 10 µV/deg C

‹ Sometimes specification given from TMIN to TMAX
‹ Using null pins can degrade offset drift

„ Vs time:
‹ Aging in analog ICs is not linear, but rather follows a square root
function (often referred to as a “drunkard’s walk”), and is typically
based on 1,000 hours, e.g. a figure of 1µV/1000 hrs gives:-
– After 1,000 hours : 1 µV
– After 4,000 hours : 2 µV
– After 9,000 hours : 3 µV

a -22-
 Where is a Precision Op Amp
Required?

Low Level Sensor Measurements

Tambient

VTC
+
Thermocouple EOS
T1
-

„ A Thermocouple is used to measure temperatures up to 2,500 deg C

„ Its full scale output averages only 40-50 millivolts
„ Average change vs temperature VTC only 20 uV/deg C
„ Amplifier EOS change @ Tambient should be << 20 µV/deg C to be
negligible

a -23-
 Where is a Precision Op Amp
Required?
Multi-Channel Data Acquisition Systems

+
Channel 1 Sensor 1

- OP AMP 1

+
Channel 32 Sensor 32

- OP AMP 32

„ Manual adjustment of trim pots for EOS in multi-channel systems both costly and
time-consuming
„ To eliminate trims, select op amp with EOS << zero scale tolerance (1/5th -1/10th )
„ Example : if zero scale tolerance is + 1 mV , select op amp with EOS 100-200 µV

a -24-
 Low Bias Current Amplifiers

Low bias current amplifiers are characterized by...

‹ Very low input bias currents : < 1 pA.
‹ Very high input impedance : > 1012 ohms.

a -25-
 Input Bias Current, IB

IB+
+
IB-
-

„ Modeled as a pair of current sources in series with the (+) and (-)
terminals
„ Difference or offset current = IB+ - IB-
„ The bias currents of simple structures generally flow in the same
direction
„ The bias currents of more complex structures (e.g., bias-
compensated, rail-rail and transimpedance) may flow in opposite
directions

a -26-
 What is Bias Current?

„ A bipolar junction transistor

(BJT) requires bias (base) current
to flow in order to operate within
Q1 Q2 its linear region.
„ A dc return path to ground is
required for bias current to flow.
IB1 IB2 „ A field effect transistor (FET) has
very low bias current (and very
high input impedance) compared
to a bipolar transistor, but still
needs a dc path.

a -27-
 Typical Values of Ib

„ BJT (Bipolar Junction Transistor):

‹ Typically 10 nA to 10 mA
‹ Varies linearly With temperature

‹ ZIN typically 100’s of Megohms

„ FET (Field Effect Transistor):

‹ Typically << 1 pA
‹ Doubles for every 10 deg C increase in
temperature
‹ ZIN typically 1012 ohms

a -28-
 The Effects of Input Bias Current
RF

RI
-

IB-
VIN
+
IB+

VOUT = - (VIN/RI + IB-) x RF

The bias current will introduce an error term depending on the

values of the bias current and the source resistance

a -29-
 Measuring Input Bias Current
S1
+
RS -
- VOUT
+
100 Ω
S2
R2

R2
VOUT = (1 + ------) x (EOS + IB+RS – IB-RS)
100

„ Rs >> 100 Ω (typ 100 kΩ to 1 GΩ)

„ Close S1 to measure IB+ and S2 to measure
IB-
„ Close both to measure EOS

a -30-
Bias Current Cancellation Methods -
External

In the circuit shown below, R3 is chosen such that R3 = R1||R2

R2
IB-
R1
-

+
R3
IB+

a -31-
Bias Current Cancellation Methods –
Internal

A bias current compensated bipolar input

stage has all of the features of the
standard bipolar input stage – low offset
voltage, low drift and low voltage noise –
but exhibits a higher current noise. In
addition, the bias currents of a
compensated input stage may flow either
in the same or in opposite directions.
The bias currents of a bipolar input stage
may also be reduced by the use of super
beta transistors, although these transistor
types have a limited frequency response
compared to standard bipolar.

a -32-
Total Error Due to Offset Voltage
and Bias Current

R2
R1 IB-

- +
+
Rp EOS IB+

R2 R2
VOUT = + EOS [1 + ------] + IB+ RP [1 + -----] + IB-R2
R1 R1

a -33-
 AC Input Coupling

RA Adding a capacitor in series

with either input to block any
+
external DC offset violates the
ground return path rule.
-
RF
RA RB

RI
+

-
Adding a resistor from the (+) RF
input to ground restores the
necessary bias current return
path to ground. RI

a -34-
Where is a Low Bias Current Op Amp
Required?
Sensing Photodetector Current

IP RF
-
Light
Source +
VOUT = IP x RF
RB

„ The photodetector produces a current proportional to its illumination due

to the light source
„ Currents can range from “dark” - femto (10-12) or nano (10-9) - amperes to
several milliamperes (10-3)

a -35-
 Equivalent Photodiode Circuit

An FET Amplifier is the Best Choice for This Type of Application

Suitable types include AD549, AD546, AD795 and AD645

RF
-

RP CP +

≈ RB

a -36-
 Equivalent Photodiode Circuit

„ RP equals photodiode resistance, typically 100-1000

Megohms

„ IP equals photodiode current, typically 1 pA - 1 nA

„ RF typically several hundred Megohms

„ RB equals bias current compensation resistance equal to

RF//RP

„ Ultra-low (<< 1 pA) bias current FET input amplifier is

required

a -37-
 Input Bias Current Cancellation
Here is an example of where the not so obvious resulted in an unusual and
surprising result for one of our customers (based on an actual call)

Conditions @ + 25 deg C RF = 1000 MΩ

VBIAS = 0.75 mV

-
AD645
IB
+
RP = 1000 MΩ IB VOUT ≈ 750 µV
RB = 500 MΩ

IB = 1.5 pA @ + 25 deg C

DC Noise Gain = 2
a -38-
 Bias Current Cancellation - a
Disaster!

Conditions @ + 85 deg C
RF = 1000 MΩ

VBIAS = 48 mV

-
AD645
IB
+
RP = 15.7 MΩ IB VOUT = 3V !
RB = 500 MΩ

IB = 96 pA @ + 85 deg C

DC Noise Gain = 64.7

a -39-
 What Happened at 85 deg C?
Customer didn’t realize
1 that RP varies inversely RF = 1000 MΩ
with temperature
VBIAS = 48 mV

-
AD645 VOUT = 3V !
IB
+
RP = 15.7 MΩ IB
RB = 500 MΩ

IB = 96 pA @ + 85 deg C
DC Noise Gain = 64.7
Customer also didn’t
The new ratio of RF / RP
3 has increased the noise 2 realize (or forgot) that IB
doubles every 10 deg C
gain to 64.7

a -40-
 Bias Current Cancellation...The
Solution

Calculate Rb = Rp//Rf for Rp @ Tmax (e.g., + 85 deg C)

RF = 1000 MΩ

VBIAS = 1.5 mV

-
AD645
IB
+
RP = 1000 MΩ VOUT = 0
IB
RB = 15.5 MΩ

IB = 1.5 pA @ + 25 deg C
DC Noise Gain = 2

a -41-
 The Solution (con’t)
Output Voltage Using Values of RB calculated for RP at 25 deg C and 85 deg C

TEMPERATURE RP CALCULATED @ RP CALCULATED @

25 DEG C 85 DEG C

25 deg C 0 mV -1.5 mV

35 deg C 1.5 mV -2.9 mV

45 deg C 9.0 mV -5.5 mV

55 deg C 42.0 mV -10.3 mV

65 deg C 180 mV -17.6 mV

75 deg C 745 mV -23.3 mV

85 deg C 3.00 V +1.34 mV

a -42-
 Low Noise Amplifiers

Low noise amplifiers are characterized by…

‹ Low voltage noise densities, en < 10 nV per root Hz
‹ Low current noise densities, in < 10 pA per root Hz

a -43-
Voltage Noise, EN and Current Noise,
IN

IN+ EN
+
IN-
-

„ An amplifier has three (3) internal noise sources:

¾ Voltage noise
¾ Current noise in the inverting input
¾ Current noise in the non-inverting input
„ Voltage noise also has a low frequency component known as
1/f noise

a -44-
 What is the Effect of Noise in an
Amplifier?

AA –– D
D
CONVERTER
CONVERTER

„ It reduces the effective dynamic range of the A/D

converter, making it difficult to measure very small AC
signals
„ Since noise is a continuous spectrum of all frequencies
from DC – infinity, it cannot be easily filtered out within the
passband of interest
„ Out of band noise, however can be filtered out using a
variety of active or passive techniques

a -45-
 Voltage Noise
The voltage noise of an op amp (en) is due to the Johnson noise of the resistors
within the op amp. The noise exhibits a flat spectrum (also called “white
noise”), which means it exhibits the same noise power in each Hz of frequency.

The Johnson Noise of a Resistor Is Expressed As, -

Enoise (rms) = (4kTRB)1/2

Where,
„ Enoise = Total Noise in µV
„ K = Boltzman’s Constant = 1.38 x 10-23
Joules/deg K
„ T =Temperature in Degrees Kelvin
„ R = Resistance in Ohms

„ B = Bandwidth in Hertz

a -46-
 1/f (Flicker) Noise
Fluctuations in the amplifier’s internal DC resistances is called 1/f or flicker noise..It
has a 1/f spectral response (also called “pink noise”), which means it exhibits equal
noise power per decade of frequency. The 1/f corner frequency is a “figure of merit”
for the amplifier – the lower the better
Noise Spectral Density

Slope = -10db/decade or – 3 db/octave

(nV per root Hz)

asymptotes

Frequency (Log Scale)

1/f
Corner frequency (typ 10 – 100 Hz)
a -47-
 Current Noise

The current noise of the op amp is due to the shot or Schottky noise,
produced by fluctuations in the DC bias current. Since electrical
current is not a smooth, continuous flow but rather a stream of
discrete electric charges, shot noise is produced by the statistical
fluctuations of the current due to the finite nature of the charge
quantum.

INOISE (RMS) = (2q x IB x B)½

where, -
„ q = electron charge = 1.6 x 10-19 coulombs

„ IB = amplifier bias current

„ B = bandwidth in Hertz

a -48-
 Popcorn Noise
Popcorn noise was a dominant noise source in the 1960’s, and was due primarily
to contamination in the IC manufacturing processes. The noise appeared as
random step changes in the offset voltage over periods of 10 milliseconds or
more, hence the term “popcorn” noise. Precision voltage references were also
susceptible to this type of noise.
Today’s advanced IC processes and clean room environments have all but
eliminated this problem.

Nominal
VOS
T
5 uV

a -49-
 Noise Spectral Density
Both the voltage noise and the current noise are specified in nV
per root Hz and pA per root Hz, respectively, to reflect their
dependence on the square root of the bandwidth, as shown in their
respective equations:

Enoise (rms) = (4kTRB)1/2

INOISE (rms) = (2q x IB x B)1/2

This allows a designer to determine the total RMS noise in a given

bandwidth, regardless of the center frequency

a -50-
Typical Input Noise Density Curves
(AD705)

a -51-
 RMS vs Peak-Peak Noise
The peak-peak noise determines the true accuracy of the measurement
system…the values are based on a Gaussian distribution…

Nominal % of the Time Noise Will Number of Events per

Peak – Peak Value Exceed Nominal Million
Peak – Peak Value

2 x rms 32% 320,000

3 x rms 13% 130,000

4 x rms 4.6% 46,000

5 x rms 1.2% 12,000

6 x rms 0.27% 2,700

6.6 x rms 0.10% 1,000

7 x rms 0.048% 480

8 x rms 0.006% 60

a -52-
 Effect of Current Noise Due to
Source Resistance
The dominant noise source for each value of R is shown in the shaded area

OP27 :
Voltage Noise = 3 nV per root Hz
- Current Noise = 1 pA per root Hz
OP-27
+ Noise (nV/rt Hz)

R
Value of R: 0 3 kΩ 300 kΩ

1. Amplifier Voltage Noise 3 3 3

2. INoise Flowing in R 0 3 300

3. Johnson Noise in R 0 7 70

a -53-
 Noise Figure
In communications applications you will sometimes hear the term “noise figure”. It
is the ratio, in db of the output noise of a “real” amplifier to the output of an “ideal”
(noiseless) amplifier with a source resistance RS (typically 50-75 ohms) across its
input.

It is only useful when comparing amplifiers under identical conditions of source

impedance and it should not be used as an op-amp specification.

+ VN2
RS N.F. = 10log10 (1 + -----------)
4kTRS
-

a -54-
 High Speed Amplifiers

High speed amplifiers are characterized by…

‹ High gain bandwidth product >10 MHz
‹ Slew rates > 50 V/µs
‹ Settling times < 100 ns

a -55-
 What is Gain Bandwidth Product
(GBP)?

Slope = 6 dB per Octave

GAIN Y1 and
dB Noise Gain = Y = 1 + RF/RI
Y2

fCL1 fCL2 LOG F

„ It is the product of the closed loop gain Y and the closed loop bandwidth
fcl
„ Y1 * fcl1 = Y2 * fcl2 = CONSTANT
„ Only applies to voltage feedback amplifiers

a -56-
 Slew Rate
Slew rate is the maximum rate at which the output of an amplifier can
change, in response to a change at its input. Slew rate and full power
bandwidth are interrelated.

+
voltage voltage

-
time time

Slew Rate
Slew Rate = VPEAK x 2 x pi x BWFP Full Power Bandwidth = ----------------------
2 x pi x VPEAK

a -57-
 Slew Rate

The following example will illustrate how slew rate does not
always equate with high speed or high bandwidth…

+
20 kHz
SINEWAVE
20 kHz
TRI-WAVE ?
2.5V p-p 5V p-p
-

Amplifier : type OP-07, gain = + 4

WHAT’S THE PROBLEM?

a -58-
 Slew Rate

+
voltage voltage

-
time time

Answer : problem is slew rate limiting in the OP-07 amplifier:

„ Maximum rate of change of sine wave occurs at zero crossing :
1.2 volts per microsecond at 20 kHz
„ Slew rate of the OP-07 amplifier : approximately 0.1 volts per
microsecond!

a -59-
 Small Signal vs Full Power
Bandwidth

„ Small Signal Bandwidth:

‹ Refers to the frequency response of an amplifier which is
not slew rate limited.
‹ Measured with small signal typically 10-20% of full scale
‹ The amplifier’s open loop poles determine the – 3dB
bandwidth for a given closed loop configuration
„ Full Power Bandwidth:
‹ Refers to the highest frequency that will not cause slew rate
limiting in an amplifier operating at its peak output voltage
‹ Measured with full scale input

a -60-
 Settling Time
Settling time is defined as the time it takes the amplifier’s output to
respond to step change in input, and come within and remain within a
defined error band as measured relative to the 50% point of the input
pulse.
ERROR
BAND

TIME

DEAD SLEW
RECOVERY TIME FINAL SETTLING
TIME TIME

SETTLING TIME

a -61-
 Differential Gain and Phase

Differential Gain and Differential Phase are 2 important specifications

found on our high speed amplifier data sheets that are required by the
video industry…

„ Composite video signals contain color or chrominance information in

the amplitude and phase of the subcarrier.
„ The amplitude of the subcarrier determines the intensity, while the
phase of the subcarrier with respect to the color burst determines the
actual color.
„ A real world amplifier will have amplitude and phase variations across
the black to white level of the video signal.

a -62-
Differential Gain and Phase (con’t)

„ Any variation of the amplitude of the color subcarrier from

black to white level is the differential gain.
„ Any variation in phase with respect to the color subcarrier from
black to white level is the differential phase.
„ The industry standard specifications for differential gain and
differential phase are as follows:
‹ Differential gain : ≤0.1%
‹ Differential phase : ≤ 0.1 degree

a -63-
 Differential Gain and Phase:
Measured Results

0.005%
0.014% PER
DIV

DIFFERENTIAL GAIN
-1V +1V
RL = 100 Ω, MODULATION: 200mV, 4.43 MHz

0.01o
0.025o PER
DIV

DIFFERENTIAL PHASE

a -64-
 What is a Transimpedance
Amplifier?
A Transimpedance amplifier (also referred to as a current feedback
amplifier) is used primarily in high speed applications. It is similar to a
voltage feedback amplifier, except for certain architectural differences that
give it its advantage (under certain conditions) over its voltage feedback
counterpart.

a -65-
 What Are The Basic Differences?

VOLTAGE FEEDBACK: TRANSIMPEDANCE:

„ Balanced input impedance on + „ Unbalanced + and - input
and - inputs impedances
„ Constant gain bandwidth product „ Bandwidth set by feedback
„ Gain set by input and feedback resistor
resistors „ Gain set by input resistor
„ Precision applications „ High speed applications

a -66-
 Voltage Feedback vs Current
Feedback Front End

IBIAS

NONINVERTING INVERTING INVERTING NONINVERTING

INPUT INPUT INPUT INPUT
-
+ - +

IBIAS

AD829 AD826

a -67-
 Input Impedance : Voltage
Feedback

Ib+ Ib-

ZDIFF „ ZCM typically 105 - 1012 ohms // 5-

25 pf
+ INPUT - INPUT
„ In most circuits, ZCM - swamped by
ZCM+ ZCM-
negative feedback

a -68-
 Input Impedance : Transimpedance

+ INPUT - INPUT
X1
• Z+ typically 105 - 1012 ohms with
Z-
very little shunt capacitance

Z+ • Z- is typically 10-100 ohms and

may be reactive (C or L)

a -69-
 Output Impedance:
The output stages of a voltage feedback amplifier (left) and transimpedance
amplifier (right) are basically the same

AD844 Simplified Schematic

a -70-
 Remember...

Voltage Feedback:
„ Bandwidth independent of value of
R2
„ gain set by ratio of R2/R1

Current Feedback:
„ Bandwidth set by value of R2

„ Ratio R2/R1 then sets the gain

a -71-
 What Is a Single Supply
Amplifier?

A single supply amplifier is an amplifier whose common mode

input range includes ground or the negative rail

a -72-
Where are Single Supply Amplifiers
Used?
Single supply amplifiers are used in portable applications requiring
very low voltage and/or power consumption

+
+VS
+

-
-VS

-
z Supply range typically 1.8V – 5.0V DC
z Power consumption typically : 5 uA – 150 uA per amplifier

a -73-
 What Are The Tradeoffs?
„ Bandwidth:
‹ Bandwidth is typically proportional to supply current. For
very low power devices (< 150 uA quiescent current)
bandwidth is limited to around 100 kHz. Since many
portable applications are transducer-based, bandwidth may
not be an important issue.
„ Noise:
‹ The noise floor of an amplifier tends to rise as quiescent
(supply) current is reduced.
„ Product selection:
‹ The selection pool of precision amplifiers tends to shrink.
„ Dynamic range :
‹ Unless a rail-rail amplifier is used, dynamic input/output
range is severely limited

a -74-
 Single Supply vs Conventional
Amplifier Input Stages

A conventional NPN stage requires 2-3V of

headroom above and below input swing.
Attempting to go to ground will shut off the
input transistors

The use of a PNP stage allows the amplifier

input to go to the negative rail (ground). The
PNP stage still needs 2-3V of headroom from
the positive supply rail, however.

a -75-
 What is a Rail-Rail Amplifier ?

A true rail-rail amplifier can swing to within a few millivolts

of their supply rails, either on the input, the output or both.

+ VS

VIN VOUT

- VS

a -76-
 Why Rail-Rail?
+VS

+ Analog to
Digital
Converter
-

„ Many new high speed A-D converters operate from single +3V to
+5V supply
„ Rail-rail amplifiers provide maximum dynamic range

a -77-
 A True Rail-Rail Input Stage
+ VS

Q3 Q4
+ VIN Q1 Q2 - VIN

- VS

a -78-
 Traditional Output Stages

+VS +VS +VS

NPN NMOS NPN

VOUT VOUT
VOUT

NPN NMOS
PNP

–VS –VS –VS

a -79-
 "Almost" Rail-to-Rail Output
Structures
In the PNP-NPN structure the output swing is limited by the saturation voltage of
the transistors. In the MOSFET structure, output swing is limited by the ON
resistance of the FETs

+VS +VS

PNP PMOS

VOUT VOUT

NMOS
NPN

–VS –VS

a -80-
 Summary

„ An Operational Amplifier is used to amplify and/or level translate (DC

shift) low level, “real world” signals from sensors, transducers, etc to a
high level signal compatible with the input of the A-D converter.
„ A Precision Amplifier has low initial offset voltage (< 100 µV) and low
drift (< 1 µ V/deg C). It is used for precision measurement of low level
transducer (sensor) signals.
„ A Low Bias Current Amplifier has very low bias current (< 1 pA) and
very high input impedance (> 1012 ohms). It is commonly used in
photodiode detector circuits.
„ High Speed Amplifiers are used in communications and video
applications.
„ Single Supply Amplifiers are often used in low power, portable and
remote applications.
„ Rail-Rail Amplifiers extend an A/D converter’s dynamic input range to
the (+) and (–) supply rails.

a -81-
 Want To Know More?

„ “Linear Design Seminar”, Analog Devices,

Spring 1995
„ Go to ADI’s website at www.analog.com and
click on “Technical Support”…you will find
numerous application notes, technical articles
and interactive design tools on operational
amplifiers

a -82-