Professional Documents
Culture Documents
25
Inserting Equation 4 into Equation 5 and solving for Replacing G1 with GTOT and G1′ with GTOT/G2′ results in
VO/VD provides the transfer function of the INA: the extended common-mode range:
VO VD 1
= G1G 2 = GTOT . (6) VCM ′ = VCM + GTOT 1 −
VD 2 G 2′
(7)
Extending the input common-mode voltage range = VCM +
VD ′
(
G G ′ −1 .
2 1 2 )
Note that V1 and V2 in Equation 4 do not represent
absolute voltages. Because VCM and VD can change their This improved common-mode range at the amplifier
polarities, the maximum voltage either output can assume output is now passed on 1:1 to the input. Applying gain to
before reaching saturation is the difference amplifier requires access to the feedback
resistor of A3 in Figure 2. A common solution uses a
V stand-alone difference amplifier, which provides access to
± V1, 2 = ± VCM + D ≤ ± VSAT .
2 the feedback resistor via a VSENSE pin. The input stage is
then realized by a dual low-noise amplifier, with external
For clarification, the following description simply ignores
resistors RF and RG being used to set the input gain.
signal polarities, and the variables refer only to magnitude
To raise the gain of a unity-gain amplifier, external resis-
values. Assuming that V1,2 and VD/2 are constant, the only
tors can be switched in series to R2. However, the internal
way to increase the input common-mode voltage from VCM
resistor values must be measured, as they can deviate by
to VCM′ is to reduce the input gain from G1 to G1′ so that
±30% from their nominal values given in the datasheet.
VD V This approach works well for moderate gain. For large
V1,2 = constant = VCM + G = VCM ′ + D G1′ .
2 1 2 gain, however, the external resistors can reach prohibitive
values, increasing noise to an undesirable level. A buffered
Solving for VCM′ yields voltage divider in the feedback path of A3 is then required.
( )
VD Resistors R3 and R4 allow a wide range of gain settings
VCM ′ = VCM + G1 − G1′ . with moderate resistor values. Voltage follower A4 pro-
2
vides low output impedance, which preserves the high
Reducing G1 reduces the range of the amplified differ- CMRR of the difference amplifier.
ential component, G1′(VD/2), thus providing an expansion
range for VCM. Standard INAs, using unity-gain difference Removing large differential DC potentials
amplifiers, have R2 = R1 and G2 = 1. The signal conditioning in analog front ends of medical
The total INA gain is then placed into the input stage, equipment, such as electrocardiographs (ECGs), presents
making G1 = GTOT. Equation 6 shows that reducing G1 from the additional design challenge of detecting small AC sig-
GTOT to G1′, while preserving GTOT, requires an increase in nals in the presence of large differential DC potentials.
difference amplifier gain from G2 = 1 to G2′ = GTOT/G1′.
Figure 2. Increasing difference amplifier gain via REXT or buffered voltage divider
R2'
R4
+
R1 R2 REXT R1 R2
V1 V1 A4
VSENSE –
R3
– VO –
R1 = R2 A3 VO R1 = R2 A3 VO
+ +
G2' = R2'/R1 G2' = 1 + R4/R3
VREF
V2 V2 VREF
R1 R2 REXT R1 R2
R2'
26
27
Figure 4. INA128 with OPA2132 providing low-pass filter and external gain stage
+ INA128
VIN– V1 R1 R2 R4
–A1 CC
40 kΩ 40 kΩ
25 kΩ
– R3
RF A3 – RFILT
RG G1 = 1 + 2RF /R G + A4 VO
RINT + CFILT
RF CINT
G2 = 1 ½ OPA2132
25 kΩ G3 = –GTOT /G1 = –R4 /R3
– 40 kΩ 40 kΩ –
A2 A5
VIN+ + V2 R1 R2 +
½ OPA2132
To return to the specified f0 of 0.05 Hz requires the • Gain-booster A4 and integrator A5 can be designed with
increase of the time constant by the factor G2, thus quickly the dual low-noise amplifier OPA2132 with an input-
–––
leading to prohibitive values for RINT and CINT. referred noise of 8 nV/√Hz.
There are two alternatives to design around this problem. • The adjustment of G1 is independent from G2 and f0,
Either (1) change the gain settings of G1, G2, and GTOT allowing the input gain to be set for maximum input
until moderate values for RINT and CINT can be found, or common-mode range.
(2) make G2 = 1 and perform the final signal boost via a • The RC values defining the integrator time constant
separate gain stage (Figure 4). now reflect the real lower-bandwidth limit, f0.
The latter approach, which is the easier one, provides
• The final gain stage A4 allows independent adjustment
the following benefits:
of any desired gain value and performs low-pass filtering
• Standard INAs with unity-gain output stages, such as of high-frequency noise.
INA128 or INA118, can be used. Both devices allow for
input gains from 1 to 10000, providing a maximum non- Single-supply applications
linearity of 0.002%. Portable ECG equipment requiring single-supply operation
can use the high-precision analog front end in Figure 5.
5V 0.0015 µF
390 kΩ +
40 kΩ
5 kΩ 1 MΩ 5V
INA326 –
(G=5) 100 Ω
390 kΩ 40 kΩ – A3
+ ½ 0.1 µF ADS8321
3.2 MΩ
750 pF OPA2335
200 kΩ 1 µF
+ VBIAS = 2.5 V
A1 10 µF
– –
½ 5V
OPA2335 A4 5V
½ + +
REF3125 100 Ω
OPA2335
20 kΩ A5
–
1 µF OPA335 0.1 µF
390 kΩ
390 kΩ –
A2
½
+ OPA2335
28
29