Appendix A: Instrumentation Amplifier Specifications

Appendix A
INSTRUMENTATION AMPLIFIER SPECIFICATIONS
To successfully apply any electronic component, a full
understanding of its specifications is required. That is to
say, the numbers contained in a data sheet are of little
value if the user does not have a clear picture of what
each specification means.
In this section, a typical monolithic instrumentation
amplifier data sheet is reviewed. Some of the more
important specifications are discussed in terms of how
they are measured and what errors they might contribute
to the overall performance of the circuit.
Table A-1 shows a portion of the data sheet for the Analog
Devices AD8221 instrumentation amplifier.
Table A-1. AD8221 Specifications1
A
Parameter
Conditions
Min
AR Grade
Typ
Max
Min
BR Grade
Typ
Max
ARM Grade
Min
Typ Max Unit
B COMMON-MODE
REJECTION RATIO (CMRR)
CMRR DC to 60 Hz with
VCM = –10 V to +10 V
1 k Source Imbalance
G = 1
80
90
80
G = 10
100
110
100
G = 100
120
130
120
G = 1000
130
140
130
VCM = –10 V to +10 V
C CMRR at 10 kHz G = 1
80
80
80
G = 10
90
100
90
G = 100
100
110
100
G = 1000
100
110
100
RTI noise = √eNI2 + (eNO/G)2
NOISE
Voltage Noise, 1 kHz
VIN+, VIN–, VREF = 0
8
8
8
Input Voltage Noise, eNI
75
75
75
Output Voltage Noise, eNO
RTI
f = 0.1 Hz to 10 Hz
G = 1
2
2
2
G = 10
0.5
0.5
0.5
G = 100 to 1000
0.25
0.25
0.25
Current Noise
f = 1 kHz
40
40
40
f = 0.1 Hz to 10 Hz
6
6
6
D VOLTAGE OFFSET2 VS = 5 V to 15 V
60
25
70
Input Offset, VOSI
86
45
135
Over Temperature
T = –40C to +85C
Average TC
0.4
0.3
0.9
VS = 5 V to 15 V
300
200
600
Output Offset, VOSO
0.66
0.45
1.00
Over Temperature
T = –40C to +85C
Average TC
6
5
9
Offset RTI vs. Supply (PSR) VS = 2.3 V to 18 V
G = 1
90
110
94
110
90
100
G = 10
110
120
114
130
100
120
G = 100
124
130
130
140
120
140
G = 1000
130
140
140
150
120
140
E INPUT CURRENT
Input Bias Current
0.5
1.5
0.2
0.4
0.5
2
2.0
1
3
Over Temperature
T = –40C to +85C
Average TC
1
1
3
Input Offset Current
0.2
0.6
0.1
0.4
0.3
1
0.8
0.6
1.5
Over Temperature
T = –40C to +85C
Average TC
1
1
3
A-1
dB
dB
dB
dB
dB
dB
dB
dB
nV/√Hz
nV/√Hz
µV p-p
µV p-p
µV p-p
fA/√Hz
pA p-p
µV
µV
µV/C
µV
mV
µV/C
dB
dB
dB
dB
nA
nA
pA/C
nA
nA
pA/C
Parameter
Conditions
Min
AR Grade
Typ
Max
H
BR Grade
Typ
Max
Min
ARM Grade
Typ
Max
k
A
V
V/V
POWER SUPPLY
18
2.3
1
0.9
1
1.2
1
Over Temperature
T = –40C to +85C
V
mA
mA
18
2.3
1
0.9
1.2
1
18
1
1.2
DYNAMIC RESPONSE
Small Signal, –3 dB
Bandwidth
G = 1 825
825
825
G = 10
562
562
562
G = 100
100
100
100
G = 1000
14.7
14.7
14.7
Settling Time 0.01%
10 V step
G = 1 to 100
10
10
10
G = 1000
80
80
80
Settling Time 0.001%
10 V step
G = 1 to 100
13
13
13
G = 1000
110
110
110
Slew Rate
G = 1
1.5
1.7
1.5
1.7
1.5
1.7
G = 5 to 100
2
2.5
2
2.5
2
2.5
I GAIN
G = 1 + (49.4 k/RG)
J Gain Range
1
1000
1
1000
1
1000
VOUT 10 V
K Gain Error
G = 1
0.03
0.02
0.1
G = 10
0.3
0.15
0.3
G = 100
0.3
0.15
0.3
G = 1000
0.3
0.15
0.3
VOUT = –10 V to +10 V
L Gain Nonlinearity
3
10
3
10
5
15
G = 1 to 10RL = 10 k
5
15
5
15
7
20
G = 100RL = 10 k
10
40
10
40
10
50
G = 1000RL = 10 k
10
95
10
95
15
100
G = 1 to 100RL = 2 k
M Gain vs. Temperature
G = 1
3
10
2
5
3
10
–50
–50
–50
G > 13
INPUT
Input Impedance
Differential
100 ||2
100 ||2
100 ||2
Common Mode
100 ||2
100 ||2
100 ||2
Input Operating
VS = 2.3 V to 5 V –VS + 1.9
+VS – 1.1 –VS + 1.9
+VS – 1.1 –VS + 1.9
+VS – 1.1
Voltage Range4
+VS – 1.2 –VS + 2.0
+VS – 1.2 –VS + 2.0
+VS – 1.2
Over Temperature
T = –40C to +85C –VS + 2.0
Input Operating
+VS – 1.2 –VS + 1.9
+VS – 1.2 –VS + 1.9
+VS – 1.2
Voltage Range
VS = 5 V to 18 V –VS + 1.9
+VS – 1.2 –VS + 2.0
+VS – 1.2 –VS + 2.0
+VS – 1.2
Over Temperature
T = –40C to +85C –VS + 2.0
N
Unit
REFERENCE INPUT
RIN
20
20
20
VIN+, VIN–, VREF = 0
50
60
50
60
50
60
IIN
+VS
–VS
+VS
–VS
+VS
Voltage Range
–VS
1  0.0001
1  0.0001
Gain to Output
1  0.0001
F Operating Range
VS = 2.3 V to 18 V 2.3
0.9
G Quiescent Current
H
Min
kHz
kHz
kHz
kHz
s
s
s
s
V/s
V/s
V/V
%
%
%
%
ppm
ppm
ppm
ppm
ppm/C
ppm/C
G ||pF
G ||pF
V
V
V
V
OUTPUTRL = 10 k
+VS – 1.2 –VS + 1.1
+VS – 1.2 –VS + 1.1
+VS – 1.2
Output Swing
VS = 2.3 V to 5 V –VS + 1.1
+VS – 1.3 –VS + 1.4
+VS – 1.3 –VS + 1.4
+VS – 1.3
Over Temperature
T = –40C to +85C –VS + 1.4
+VS – 1.4 –VS + 1.2
+VS – 1.4 –VS + 1.2
+VS – 1.4
Output Swing
VS = 5 V to 18 V –VS + 1.2
+VS – 1.5 –VS + 1.6
+VS – 1.5 –VS + 1.6
+VS – 1.5
Over Temperature
T = –40C to +85C –VS + 1.6
Short-Circuit Current
18
18
18
V
V
V
V
mA
TEMPERATURE RANGE
Specified Performance –40
+85
–40
+85
–40
+85
–40
+125
–40
+125
–40
+125
Operational4
°C
°C
NOTES
1
VS = 15 V, VREF = 0 V, TA = +25C, G = 1, RL = 2 k, unless otherwise noted.
2
Total RTI VOS = (VOSI) + (VOSO/G).
3
Does not include the effects of external resistor RG.
4
One input grounded. G = 1.
A-2
A statement at the top of the data sheet explains that
the listed specifications are typically @ TA = 25C,
VS = 15 V, and RL = 10 k, unless otherwise noted.
This tells the user that these are the normal operating
conditions under which the device is tested. Deviations
from these conditions might degrade (or improve) performance. For situations where deviations from the normal
conditions (such as a change in temperature) are likely,
the significant effects are usually indicated within the
specs.The statement at the top of the specifications table
also tells us what all numbers are unless noted; typical
is used to state that the manufacturer’s characterization
process has shown a number to be average; however,
individual devices may vary.
Instrumentation amplifiers designed for true rail-to-rail
operation have a few critical specifications that need to
be considered.Their input voltage range should allow the
in-amp to accept input signal levels that are close to the
power supply or ground. Their output swing should be
within 0.1 V of the supply line or ground. In contrast, a
typical dual-supply in-amp can swing only within 2 V or
more of the supply or ground. In 5 V single-supply data
acquisition systems, an extended output swing is vital
because it allows the full input range of the ADC to be
used, providing high resolution.
(B) Common-Mode Rejection
Common-mode rejection is a measure of the change in
output voltage when the same voltage is applied to both
inputs. CMR is normally specified as input, which allows
for in-amp gain. As the gain is increased, there will be a
higher output voltage for the same common-mode input
voltage. These specifications may be given for either a
full range input voltage change or for a specified source
imbalance in ohms.
Common-mode rejection ratio is a ratio expression,
while common-mode rejection is the logarithm of
that ratio. Both specifications are normally referred to
output (RTO).
That is,
CMRR =
Change inOutputVoltage
Change in Common-Mode InputVoltage
While
CMR = 20 Log10 CMRR
For example, a CMRR of 10,000 corresponds to a CMR
of 80 dB. For most in-amps, the CMR increases with
gain because most designs have a front-end configuration that rejects common-mode signals while amplifying
differential (i.e., signal) voltages.
Common-mode rejection is usually specified for a full
range common-mode voltage change at a given frequency,
and a specified imbalance of source impedance (e.g., l k
source unbalance, at 60 Hz).
(C) AC Common-Mode Rejection
As might be expected, an in-amp’s common-mode rejection does vary with frequency. Usually, CMR is specified
at dc or at very low input frequencies. At higher gains, an
in-amp’s bandwidth does decrease, lowering its gain and
introducing additional phase shift in its input stage.
Since any imbalance in phase shift in the differential input
stage will show up as a common-mode error, ac CMRR
will usually decrease with frequency. Figure A-1 shows
the CMR vs. frequency of the AD8221.
160
140
120
CMR (dB)
(A) Specifications (Conditions)
100
GAIN = 1000
GAIN = 100
GAIN = 1000
GAIN = 10
GAIN = 1
80
GAIN = 10
GAIN = 100
60
40
0.1
1
10
100
1k
10k
100k
1M
FREQUENCY (Hz)
Figure A-1. AD8221 CMR vs. frequency.
(D) Voltage Offset
Voltage offset specifications are often considered a
figure of merit for instrumentation amplifiers. While
any initial offset may be adjusted to zero through the
use of hardware or software, shifts in offset voltage due
to temperature variations are more difficult to correct.
Intelligent systems using a microprocessor can use a
temperature reference and calibration data to correct
for this, but there are many small signal, high gain applications that do not have this capability.
A-3
Voltage offset and drift comprise four separate error
definitions: room temperature (25C), input and output,
offset, and offset drift over temperature referred to both
input and output.
An in-amp should be regarded as a 2-stage amplifier with
both an input and an output section. Each section has
its own error sources. Because the errors of the output
section are multiplied by a fixed gain (usually 2), this
section is often the principal error source at low circuit
gains. When the in-amp is operating at higher gains, the
gain of the input stage is increased. As the gain is raised,
errors contributed by the input section are multiplied,
while output errors are reduced. Thus, at high gains, the
input stage errors dominate.
Input errors are those contributed by the input stage
alone; output errors are those due to the output section.
Input-related specifications are often combined and
classified together as referred to input (RTI) errors, while
all output-related specifications are considered referred to
output (RTO) errors. It is important to understand that
although these two specifications often provide numbers
that are not the same, either error term is correct because
each defines the total error in a different way.
For a given gain, an in-amp’s input and output errors
can be calculated using the following formulas:
Total Error, RTI = Input Error + (Output Error/Gain)
Total Error, RTO = (Gain  Input Error) + Output Error
Sometimes the specification page will list an error term as
RTI or RTO for a specified gain. In other cases, it is up
to the user to calculate the error for the desired gain.
As an example, the total voltage offset error of the
AD620A in-amp when it is operating at a gain of 10
can be calculated using the individual errors listed on
its specifications page. The (typical) input offset of the
AD620 (VOSI) is listed as 30 µV. Its output offset (VOSO)
is listed as 400 µV. The total voltage offset referred to
input (RTI) is equal to
Total RTI Error = VOSI + (VOSO/G) = 30 V +
(400 V/10) = 30 V + 40 V = 70 V
(E) Input Bias and Offset Currents
Input bias currents are those currents flowing into or
out of the input terminals of the in-amp. In-amps using
FET input stages have lower room temperature bias currents than their bipolar cousins, but FET input currents
double approximately every 11C. Input bias currents
can be considered a source of voltage offset error (i.e.,
input current flowing through a source resistance causes
a voltage offset). Any change in bias current is usually of
more concern than the magnitude of the bias current.
Input offset current is the difference between the two input
bias currents. It leads to offset errors in in-amps when source
resistances in the two input terminals are unequal.
Although instrumentation amplifiers have differential
inputs, there must be a return path for their bias currents to flow to common (ground).
If this return path is not provided, the bases (or gates)
of the input devices are left floating (unconnected), and
the in-amp’s output will rapidly drift either to common
or to the supply.
Therefore, when amplifying floating input sources such
as transformers (those without a center tap ground connection), ungrounded thermocouples, or any ac-coupled
input sources, there must still be a dc path from each
input to ground. A high value resistor of 1 M to 10 M
connected between each input and ground will normally
be all that is needed to correct this condition.
(F) Operating Voltage Range
A single-supply in-amp should have the same overall
operating voltage range whether it is using single or
dual supplies. That is, a single-supply in-amp, which is
specified to operate with dual-supply voltages from 1 V
to 18 V, should also operate over a 2 V to 36 V range
with a single supply, but this may not always be the case.
In fact, some in-amps, such as the AD623, will operate
to even lower equivalent voltage levels in single-supply
mode than with a dual-supply mode. For this reason, it
is always best to check the data sheet specifications.
(G) Quiescent Supply Current
The total voltage offset referred to the output (RTO)
is equal to
Total Offset Error RTO = (G (VOSI)) + VOSO =
(10 (30 V)) + 400 V = 700 V.
Note that the RTO error is 10 times greater in value
than the RTI error. Logically, it should be, because at a
gain of 10, the error at the output of the in-amp should
be 10 times the error at the input.
This specifies the quiescent or nonsignal power supply
current consumed by an in-amp within a specified
operating voltage range.
With the increasing number of battery-powered applications, device power consumption becomes a critical
design factor. Products such as the AD627 have a very
low quiescent current consumption of only 60 A, which
at 5 V is only 0.3 mW. Compare this power level to that
of an older, vintage dual-supply product, such as the
AD526. That device draws 14 mA with a 15 V supply
A-4
(30 V total) for a whopping 420 mW, 1400 times the
power consumption of the AD627. The implications for
battery life are dramatic.
With the introduction of products such as the AD627,
very impressive overall performance is achieved while only
microamps of supply current are consumed. Of course,
some trade-offs are usually necessary, so micropower
in-amps tend to have lower bandwidth and higher noise
than full power devices. The ability to operate rail-to-rail
from a single-supply voltage is an essential feature of any
micropower in-amp.
(H) Settling Time
Settling time is defined as the length of time required
for the output voltage to approach, and remain within, a
certain tolerance of its final value. It is usually specified
for a fast full-scale input step and includes output slewing time. Since several factors contribute to the overall
settling time, fast settling to 0.1% does not necessarily
mean proportionally fast settling to 0.01%. In addition,
settling time is not necessarily a function of gain. Some
of the contributing factors to long settling times include
slew rate limiting, underdamping (ringing), and thermal
gradients (long tails).
(I) Gain
These specifications relate to the transfer function of the
device. The product’s gain equation is normally listed at
the beginning of the specifications page.
The gain equation of the AD8221 is
Gain =
49,400 Ω
+1
RG
To select an RG for a given gain, solve the following
equation for RG:
RG =
49, 400 Ω
G −1
The following are samples of calculated resistance for
some common gains:
G = 1: RG =  (open circuit)
G = 9.998: RG = 5.49 k
G = 100: RG = 499 
G = 991: RG = 49.9 
Note that there will be a gain error if the standard resistance values are different from those calculated. In
addition, the tolerance of the resistors used (normally 1%
metal film) will also affect accuracy. There also will be
gain drift, typically 50 ppm/C to 100 ppm/C, if standard
resistors are used. Of course, the user must provide a very
clean (low leakage) circuit board to realize an accurate
gain of 1, since even a 200 M leakage resistance will
cause a gain error of 0.2%.
Normal metal film resistors are within 1% of their
stated value, which means that any two resistors could
be as much as 2% different in value from one another.
Thin film resistors in monolithic integrated circuits
have an absolute tolerance of only 20%. The matching
between resistors on the same chip, however, can be
excellent —typically better than 0.1%—and resistors
on the same chip will track each other thermally, so
gain drift over temperature is greatly reduced.
(J) Gain Range
Often specified as having a gain range of 1 to 1000, many
instrumentation amplifiers will often operate at higher
gains than 1000, but the manufacturer will not promise
a specific level of performance.
(K) Gain Error
In practice, as the gain resistor becomes increasingly
smaller, any errors due to the resistance of the metal runs
and bond wires inside the IC package become significant.
These errors, along with an increase in noise and drift,
may make higher gains impractical.
In 3-op amp and in-amp designs, both gain accuracy
and gain drift may suffer because the external resistor
does not exactly ratio match the IC’s internal resistors.
Moreover, the resistor chosen is usually the closest 1%
metal film value commonly available, rather than the
calculated resistance value; so this adds an additional
gain error. Some in-amps, such as the AD8230, use two
resistors to set gain. Assuming that gain is set solely by
the ratio of these two resistors in the IC, this can provide
potentially significant improvement in both gain accuracy
and drift. The best possible performance is provided by
monolithic in-amps that have all their resistors internal
to the IC, such as the AD621.
The number provided for this specification describes
maximum deviation from the gain equation. Monolithic
in-amps, such as the AD8221, have very low factorytrimmed gain errors. Although externally connected
gain networks allow the user to set the gain exactly, the
temperature coefficients of these external resistors and
the temperature differences between individual resistors
within the network all contribute to the circuit’s overall
gain error.
A-5
If the data eventually is digitized and fed to an intelligent
system (such as a microprocessor), it may be possible to
correct for gain errors by measuring a known reference
voltage and then multiplying by a constant.
(L) Nonlinearity
This makes trimming much easier to implement but may
result in nonlinearity errors of up to twice those attained
using the best straight line technique. This worst-case
error will occur when the transfer function is bowed in
one direction only.
Nonlinearity is defined as the deviation from a straight
line on the plot of an in-amp’s output voltage vs. input
voltage. Figure A-2 shows the transfer function of a
device with exaggerated nonlinearity.
Most linear devices, such as instrumentation amplifiers,
are specified for best straight line linearity. This needs
to be considered when evaluating the error budget for a
particular application.
The magnitude of this error is equal to
Regardless of the method used to specify nonlinearity,
the errors thus created are irreducible. That is to say,
these errors are neither fixed nor proportional to input
or output voltage and, therefore, cannot be reduced by
external adjustment.
Nonlinearity =
Actual Output – Calculated Output
Rated Full Scale Output Range
This deviation can be specified relative to any straight
line or to a specific straight line.There are two commonly
used methods of specifying this ideal straight line relative
to the performance of the device.
VOUT
GAIN
� + MAX
IDEAL
(STRAIGHT LINE)
ACTUAL
RESPONSE
–VIN FULL SCALE
VIN
� – MAX
(M) Gain vs. Temperature
These numbers provide both maximum and typical
deviations from the gain equation as a function of temperature. As stated in the Gain Error section (K), the TC
of an external gain resistor will never exactly match that
of other resistors within the IC package. Therefore, the
best performance over temperature is usually achieved by
in-amps using all internal gain resistors. Gain drift error
can be subtracted out in software by using a temperature
reference and calibration data.
(N) Key Specifications for Single-Supply In-Amps
+VIN FULL SCALE
There are some specifications that apply to single-supply
(i.e., rail-to-rail) in-amp products, which are of great
importance to designers powering in-amps from low
voltage, single-supply voltages.
� + MAX > � – MAX
� + MAX + � – MAX = K
Input and Output Voltage Swing
Figure A-2. Transfer function illustrating
exaggerated nonlinearity.
The best straight line method of defining nonlinearity
consists of measuring the peak positive and the peak
negative deviation and then adjusting the gain and offset
of the in-amp so that these maximum positive and negative
errors are equal. For monolithic in-amps, this is usually
accomplished by laser-trimming thin film resistors or
by other means. The best straight line method provides
impressive specifications, but it is much more difficult
to perform. The entire output signal range needs to be
examined before trimming to determine the maximum
positive and negative deviations.
The endpoint method of specifying nonlinearity requires
that any offset and/or gain calibrations are performed at
the minimum and maximum extremes of the output range.
Usually offset is trimmed at a very low output level, while
scale factor is trimmed near the maximum output level.
A single-supply in-amp needs to be able to handle
input voltages that are very close to the supply and
ground. In a typical dual-supply in-amp, the input
(and output) voltage range is within about 2 V of the
supply or ground. This becomes a real problem when
the device is powered from a 5 V supply, or can be
especially difficult when using the new 3.3 V standard.
A standard in-amp operating from a 5 V single-supply
line has only about 1 V of headroom remaining; with
a 3.3 V supply, it has virtually none.
Fortunately, a decent single-supply in-amp, such as the
AD627, will allow an output swing within 100 mV of
the supply and ground. The input level is somewhat less,
within 100 mV of ground and 1 V of the supply rail. In
critical applications, the reference terminal of the in-amp
can be moved off center to allow a symmetrical input
voltage range.
A-6