IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 3, MARCH 2003 391
AC-Coupled Front-End for
Biopotential Measurements
Enrique Mario Spinelli
3
, Student Member, IEEE, Ramon Pallàs-Areny, Fellow, IEEE, and
Miguel Angel Mayosky, Senior Member, IEEE
Abstract—AC coupling is essential in biopotential mea-
surements. Electrode offset potentials can be several orders of
magnitude larger than the amplitudes of the biological signals of
interest, thus limiting the admissible gain of a dc-coupled front
end to prevent amplifier saturation. A high-gain input stage needs
ac input coupling. This can be achieved by series capacitors, but
in order to provide a bias path, grounded resistors are usually in-
cluded, which degrade the common mode rejection ratio (CMRR).
This paper proposes a novel balanced input ac-coupling network
that provides a bias path without any connection to ground, thus
resulting in a high CMRR. The circuit being passive, it does not
limit the differential dc input voltage. Furthermore, differential
signals are ac coupled, whereas common-mode voltages are dc
coupled, thus allowing the closed-loop control of the dc common
mode voltage by means of a driven-right-leg circuit. This makes
the circuit compatible with common-mode dc shifting strategies
intended for single-supply biopotential amplifiers. The proposed
circuit allows the implementation of high-gain biopotential ampli-
fiers with a reduced number of parts, thus resulting in low power
consumption. An electrocardiogram amplifier built according to
the proposed design achieves a CMRR of 123 dB at 50 Hz.
Index Terms—AC coupling, biopotential amplifiers, electrode
offset potential.
I. INTRODUCTION
A
COMMON front-end in biopotential measurements is a
dc-coupled fully differential amplifier followed by a dif-
ference amplifier, as in the classical three op-amp instrumen-
tation amplifier. If the input stage has a large gain, this circuit
achieves a high common mode rejection ratio (CMRR) without
any trimmings [1]. Furthermore, the equivalent input noise only
depends on the two op-amps constituting the fully differential
amplifier. These are coveted features in biopotential amplifiers,
but due to electrode offset voltages, the overall gain is limited to
moderate values [2]. This situation is even worse in low-voltage
applications such as battery-powered amplifiers. Moreover, as
the first-stage gain is reduced, subsequent stages are needed to
Manuscript received February 7, 2002; revised November 8, 2002. Asterisk
indicates corresponding author.
E. M. Spinelli is with the Laboratorio de Electrónica Industrial, Control e In-
strumentación (LEICI), Departamento de Electrotecnia, Universidad Nacional
de La Plata (UNLP), and also with the Comisión de Investigaciones Científicas
de la Provincia de Buenos Aires (CICPBA) CC 91, 1900 La Plata, Argentina
(spinelli@ing.unlp.edu.ar)
R. Pallàs-Areny is with the Departament d’Enginyeria Electrònica, Univer-
sitat Politècnica de Catalunya (UPC) EPS Castelldefels, 08860 Castelldefels
(Barcelona), Spain.
M. A.Mayoskyiswith the Laboratorio de Electrónica Industrial, Control e In-
strumentación (LEICI), Departamento de Electrotecnia, Universidad Nacional
de La Plata (UNLP), and also with the Comisión de Investigaciones Científicas
de la Provincia de Buenos Aires (CICPBA), CC 91, 1900 La Plata, Argentina.
Digital Object Identifier 10.1109/TBME.2003.808826
Fig. 1. (a) Typical circuit for balanced ac coupling. (b) Proposed ac-coupling
circuit without any grounding resistor.
obtain a high gain, thus increasing the number of components
and, hence, power consumption.
A high-gain front-end amplifier for biopotentials needs
input ac coupling. The simplest ac-coupling technique is a
passive high-pass filter in front of a dc amplifier. Because
biopotential amplifiers are usually differential, that filter must
be differential as well. Fig. 1(a) shows a typical differential
filter for ac coupling [3]. This circuit is simple and suits
low-power applications, but resistor
reduces the input
common mode impedance
, which degrades the effective
CMRR due to the potential divider effect [4]. The CMRR of
this coupling network depends on component tolerance, it is
strongly degraded by electrode’s impedance unbalances and
decreases for increasing frequency.
For the circuit of Fig. 1(a), both CMRR and
increase with
. However, cannot be spared (infinite value), because the
amplifier needs a bias path. The effective value of grounded
resistors can be increased by bootstraping [5], but this solution
amplifies op amps’ offset more than the signal, hence, limiting
the admissible gain of the front end.
This paper presents a simple ac-coupled front end that does
not require anygrounded resistor, thus achievinga large CMRR.
II. P
ROPOSED CIRCUIT
The proposed front end includes the novel, balanced ac-
coupling network shown in Fig. 1(b). This circuit provides ac
coupling for differential signals and a dc path for amplifier bias
currents, which drain to ground through the third (common)
biopotential electrode. Hence, this solution cannot be applied
to two-electrode amplifiers. Because the input network is not
grounded, if a common mode input voltage is applied, no
currents flow through the network (there is not any path for
0018-9294/03$17.00 © 2003 IEEE
392 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 3, MARCH 2003
Fig. 2. Proposed amplifier circuit, which has two ac-coupling stages, a passive
input stage, and an active dc suppression circuit.
common mode currents), so that all network’s nodes achieve
the same potential. This absence of potential difference due
to common mode inputs implies an ideally infinite CMRR
regardless of component tolerances. In practice, however, there
are some grounded impedances not included in the model
(i.e. op amps’ input capacitances), and because of unbalanced
electrode impedances, the CMRR becomes finite. Nevertheless,
this limitation is not attributable to the ac-coupling network.
If
(see the Appendix), the transfer func-
tion of the ac-coupling network is
(1)
that corresponds to a first-order high-pass filter. If the time con-
stants are not matched, the circuit exhibits two poles and two
zeros. Nevertheless, using passive components with a reason-
able tolerance yields a transfer function quite close to (1) (see
the Appendix). Anyway, that mismatch does not degrade the
CMRR.
Because the transfer function (1) does not depend on
, one
design criterion is to select it as high as practical in order to
avoid loading effects on the input signal. Alternatively, selecting
simplifies the design. According to [7, Sec.
4.2.3.2], we need an input impedance larger than 2.44 M
at
10 Hz.
M largely fulfills this requirement.
The circuit in Fig. 1(b) removes dc input voltages, but the op
amps input offset voltages are amplified as input signals and can
significantly reduce the output dynamic range. For example, a
1-mV difference in input offset voltage and a gain of 1000 yield
1 V at the amplifier output. Moreover, the thermal noise of
and from dc to the corner frequency , and
op amps input voltage noise, including
noise, will be am-
plified. To remove offset voltage and reduce
noise, the am-
plifier itself must reject low frequencies. Fig. 2 shows a solution
that uses an integrator in a feedback loop around the difference
amplifier [3], [6].
The system has two ac-coupled stages: the front differential
ac-coupling network and the high-pass differenceamplifier. The
overall transfer function is
(2)
where
and . The firstfactor in (2)
corresponds to the passive ac-coupling network and the second
factor corresponds to the amplifier and dc restoration circuits.
A. Design Example: Electrocardiogram (ECG) Amplifier
1) Low-Frequency Response: The low-frequency behavior
of the proposed amplifier is defined by
and . These time
constants are designed to obtain the desired transient response,
which is usually given in terms of responses to rectangular or
triangular pulses. For example, a 60
V s impulse (e.g., a rect-
angular pulse of 1-mV amplitude and 60-ms duration) shall not
produce an offset on the electrocardiogram (ECG) record from
the isoelectric line greater than 20
V. The proposed circuit in-
cludes two cascaded ac stages both of which contribute to that
offset, whose amplitude will be
ms
mV
(3)
A possible choice to achieve
ms Vis s
and
s, that yields ms V.
2) High-Frequency Response: The ECG amplifier must
include a low-pass filter for bandwidth limiting. A second-order
filter with a double pole (critical damping, no overshoot in
the step response) is a good choice because it will not change
the previously calculated overshoot. Otherwise, its transient
response could increment
ms , thus leading to an iterative
design process. A double pole at 150 Hz, yields attenuations of
3 dB at 100 Hz and 6 dB at 150 Hz, which fulfill the require-
ments in [7]. The cutoff frequency was selected at 156 Hz. The
complete transfer function is then
(4)
which, with the designed time-constant values, becomes
(5)
The nominal amplifier gain was 1001, obtained with
k and .
B. Operational-Amplifier Considerations
Because the amplifier concentrates its gain in the first
stage, its equivalent input noise is determined by the op
amps composing the input stage. Low-noise applications
[i.e. high-resolution ECG, electroencephalogram (EEG)]
require low noise input op amps. In battery-powered devices,
low-power rail-to-rail op amps are an attractive choice. The
proposed design has been implemented using the low-noise
rail-to-rail TLC2274.
SPINELLI et al.: AC-COUPLED FRONT-END FOR BIOPOTENTIAL MEASUREMENTS 393
Fig. 3. Single-supply ECG amplifier based on the proposed circuit.
Fig. 4. Response of the ECG amplifier in Fig. 3 to a 1-mV, 60-ms rectangular
pulse. The insert showsthat the offset from the isoelectric line is less than 20
V.
III. EXPERIMENTAL RESULTS
Fig. 3 shows the complete circuit implemented, which
includes a driven-right-leg (DRL) circuit. The reference voltage
connected to the noninverting input of the op amps of the
DRL and the dc restoration circuits are positive to enable
single-supply operation [8]. This strategy for single-supply
operation is possible because the proposed coupling network
in Fig. 1(b) provides ac coupling for differential signals but dc
coupling for common-mode signals.
Fig. 4 shows the pulse response of the amplifier and Fig. 5
shows the frequency response. Both experimental results agree
with the predictions by (3) and (5). There is no overshoot and
the undershoot is less than 20
V, as required. The relative am-
Fig. 5. Frequency response of the ECG amplifier in Fig. 3. The analytic
expression is in dashed line and experimental data with markers.
plitude response for sinusoidal inputs falls within the tolerance
bands defined in [7, Sec. 4.2.7.3].
Fig. 6 shows the test circuit for the CMRR, which includes
impedance imbalance and cable capacitance (
200 pF). The
overall measurement circuit was shielded in order to avoid
power line interference external to the test and the measured
CMRR at 50 Hz was 123 dB, which exceeds the 60-dB
requirement in [7, Sec. 4.2.3.4].
IV. C
ONCLUSION
The simple, novel, ac-coupled front end for biopotential mea-
surements in Fig. 2 is a fully differential passive-coupling net-
work that does not include any grounded resistor, hence, re-
394 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 50, NO. 3, MARCH 2003
(6)
Fig. 6. Circuit to test the CMMR.
Fig. 7. Proposed fully differential ac-coupling network that does not include
any grounded resistor.
sulting in a high CMRR. The proposed network enables the de-
sign of a high gain for the input stage of biopotential measure-
ment systems, thus leading to implementations with a reduced
number of stages, which are particularly convenient for low-
power applications. Because the common-mode voltage is dc
coupled, the proposed circuit also suits single-supply operation.
A single-supply ECG amplifier with a gain of 1001, built ac-
cording to the design rules proposed and tested for transient and
frequency response, and CMRR, fulfilled the requirements in
[7], including a CMRR of 123 dB at 50 Hz.
A
PPENDIX
ANALYSIS OF THE PROPOSED AC COUPLING NETWORK
Fig. 7 shows an alternative drawing of the proposed passive
ac-coupling network in Fig. 2. Because it is a fully differen-
tial circuit, differential and common-mode voltages define four
transfer functions. The main transfer function
is the quo-
tient between the Laplace transforms of the differential output
voltage
and the differential input voltage . Circuit anal-
ysis yields (6), as shownat the top of the page, where
,
, , and .If , and
, reduces to
(7)
which is the same as (1). The zero-pole cancellation leading to
(7) is not very sensitive to the matching condition. Considering
unmatched passive components
(8)
we will have
and and the transfer function (6)
will display two zeros and two poles
(9)
whose location depends on component values. A first-order ap-
proximation yields
(10)
which shows that variations in
, , and with respect to
, , and , make z and to movein the same direction, so
that there is still the pole-zero cancellation. Nevertheless, (10)
is a first-order approximation and the cancellation depends on
second-order effects. For example, if
M and
F, the singularities are
(11)
Assuming a 5% tolerance,
M , M ,
F, and the zero and the poles move to
(12)
The variation in z
, , and are around 5%, as predicted by
(10), whereas z
is less than 0.5% apart from .
R
EFERENCES
[1] R. Pallàs-Areny and J. G. Webster, “Common mode rejection ratio in dif-
ferential amplifiers,” IEEE Trans. Instrum. Meas., vol. 40, pp. 669–676,
Aug. 1991.
[2] A. C. Metting Van Rijn, A. Peper, and C. A Grimbergen, “Amplifiers
for bioelectric events: a design with a minimal number of parts,” Med.
Bio. Eng. Comput., vol. 32, pp. 305–310, May 1994.
[3] R. Pallàs-Areny and J. G. Webster, Analog Signal Processing.New
York: Wiley, 1999.
[4] J. C. Huhta and J. G. Webster, “60-Hz interference in electrocardio-
graphy,” IEEE Trans. Biomed. Eng., vol. BME-20, pp. 91–101, Mar.
1973.
[5] M. J. Burke and D. T. Gleeson, “A micropower dry-electrode ECG
preamplifier,” IEEE Trans. Biomed. Eng., vol. 47, pp. 155–162, Feb.
2000.
SPINELLI et al.: AC-COUPLED FRONT-END FOR BIOPOTENTIAL MEASUREMENTS 395
[6] H. W. Smit, K. Verton, and C. A. Grimbergen, “A low-cost multichannel
preamplifier for physiological signals,” IEEE Trans. Biomed. Eng., vol.
BME-34, pp. 307–310, Apr. 1987.
[7] American National Standard ANSI/AAMI EC38:1998, Ambulatory
Electrocardiographs. Arlington, VA: Association for the Advance-
ment of Medical Instrumentation, 1999.
[8] E. M. Spinelli, N. H. Martinez, and M. A. Mayosky, “A single-supply
biopotential amplifier,” Med. Eng. Phys., vol. 23, pp. 235–238, July
2001.
Enrique Mario Spinelli (S’02) was born in
Balcarce, Argentina, in 1964. He received the
Engineer and the M.S. degrees, both in electronics,
from the Universidad Nacional de La Plata (UNLP),
Argentina, in 1989 and 2000, respectively. He is
currently working towards the Ph.D. degree at
the Facultad de Ingeniería, Universidad Nacional
de La Plata.
Since 1990, he has been with the Industrial Elec-
tronics, Control, and Instrumentation Laboratory
(LEICI), Universidad Nacional de La Plata, working
on scientific instrumentation. He is currently a Professor of control systems
at the Facultad de Ingeniería, UNLP. His current interests are biomedical
instrumentation and brain control interfaces.
Ramon Pallàs-Areny (M’81–SM’88–F’98) re-
ceived the Ingeniero Industrial and Doctor Ingeniero
Industrial degrees from the Technical University of
Catalonia (UPC), Barcelona, Spain, in 1975, and
1982, respectively.
He is currently a Professor of electronic engi-
neering at the same university, where he teaches
courses and does research in medical and electronic
instrumentation. He is the author of several books
on instrumentation in Spanish and Catalan, the latest
one being Sensors and Interfaces, Solved Problems
(Barcelona, Spain: Edicions UPC, 1999). He is also a coauthor (with J. G.
Webster) of Sensors and Signal Conditioning, 2nd ed. (New York: Wiley, 2001)
and Analog Signal Processing (New York: Wiley, 1999).
Miguel Angel Mayosky (M’97–SM’98) was born in
La Plata, Argentina, in 1960. He received the Engi-
neer on Electronics degree (First Class award) from
the University of La Plata, La Plata, Argentina, and
the Ph.D. degree in computer science from the Au-
tonomous University of Barcelona, Spain, in 1983
and 1990, respectively.
He is currently a Full Professor of automatic con-
trol systems at the School of Engineering, the Univer-
sity of La Plata, and also a Member of the Research
Staff of the Buenos Aires Scientific Research Com-
mission (CICpBA), Buenos Aires, Argentina. His research activities involve
real-time data acquisition and control systems, neural networks, and embedded
computer architectures.
