JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL. 19:7-12, 2009©
An Improved AC-amplifier for Electrophysiology
Nikola Jorgovanović, Dubravka Bojanić, Vojin Ilić and Darko Stanišić
Abstract— We present the design, simulation and test
results of a new AC amplifier for electrophysiological
measurements based on a three op-amp instrumentation
amplifier (IA). The design target was to increase the common
mode rejection ratio (CMRR), thereby improving the quality
of the recorded physiological signals in a noisy environment.
The new amplifier actively suppresses the DC component of
the differential signal and actively reduces the common mode
signal in the first stage of the IA. These functions increase the
dynamic range of the amplifier’s first stage of the differential
signal. The next step was the realization of the amplifier in a
single chip technology. The design and tests of the new AC
amplifier with a differential gain of 79.2 dB, a CMRR of 130
dB at 50 Hz, a high-pass cutoff frequency at 0.01 Hz and
common mode reduction in the first stage of the 49.8 dB are
presented in this paper.
Index Terms — AC-amplifier; electrophysiological signals;
CMRR; DC offset suppression
T
I. INTRODUCTION
HE non-invasive recordings of electrophysiological
signals are of great importance for assessing the
function of peripheral nerves, muscles, cortical activity and
the heart. The signals of interest are very small when
compared to the noise (e.g., power line interference), the
base line drift and instability, different artifacts and
electrode impedance variability [1], [2], [3] and [4]. The
standard DC three op-amp instrumentation amplifier (IA)
circuit is one of the most popular basic configurations used
for electrophysiological amplifiers. This topology provides
high input impedance and a high common mode rejection
ratio (CMRR). High CMRR can be obtained by high gain
at the first stage of the IA and well-matched resistors in the
second stage of the IA [5]. The gain at the first stage of the
DC IA needs to be limited to prevent saturation caused by
offset potentials of electrodes.
The use of an AC amplifier instead of the DC IA
eliminates the offset. An AC amplifier based on wellknown modifications to the classic three op-amp IA
topology is achievable by various techniques [6], [7], [8],
[9] and [10]. However, the techniques suggested reduce
the performance of the DC IA.
N. Jorgovanović is with University of Novi Sad, Faculty of
Engineering, Novi Sad, Serbia, Fax: +381 21 458873, Phone: +381 21
4852452, e-mail: nikolaj@uns.ac.rs
D. Bojanić is with University of Novi Sad, Faculty of Engineering,
Novi Sad, Serbia Fax: +381 21 458873, Phone: +381 21 4852446, email: dbojanic@uns.ac.rs
V. Ilić is with University of Novi Sad, Faculty of Engineering, Novi
Sad, Serbia, Fax: +381 21 458873 , Phone: +381 21 4852446, e-mail:
vojin@uns.ac.rs
D. Stanišić is with University of Novi Sad, Faculty of Engineering,
Novi Sad, Serbia, Fax: +381 21 458873, Phone: +381 21 4852452 , email: darkos@uns.ac.rs
DOI: 10.2298/JAC0901007J
Most new electrophysiological amplifier designs use
active DC suppression. This technique relates to
subtraction of the integrated IA output (amplified DC
component of the input signal) from the original input
signal (sum of the signal of interest and undesirable DC
component). Subtraction of these two signals should not
degrade the amplifier’s balanced structure or its CMRR.
The optimal solution to using this concept needs to
addresses how and when to perform the subtraction. Some
solutions for using the subtraction method are suggested
elsewhere, [11], [12], [13], [14], [15] and [16], indicating
excellent DC component suppression.
A rarely discussed factor that limits the first stage gain
of the IA and global CMRR is the dynamic range occupied
by the common mode signal. An available dynamic range
is extremely important for battery-powered devices with
low voltage power supplies. The useful dynamic range of
the first stage could be increased by additional attenuation
of the common mode signal.
We suggest a new subtraction method for the input and
the integrated output signals. Subtraction is performed in
the first stage of the IA to preserve high input impedance
and to ensure high first stage gain and high CMRR. This
new technique reduces the common mode signal in the first
stage of the amplifier to an insignificant level, thereby
practically providing full dynamic range of the amplifier’s
first stage. Moreover, common mode reduction is
performed locally in the first stage of the amplifier without
involvement of the third electrode. Therefore, the new
design can be applied with only two electrodes in the
configuration [17].
II. CIRCUIT DESIGN AND SIMULATION OF
ELECTROPHYSIOLOGICAL AMPLIFIER
A. The Novelty of the New AC Amplifier
Active suppression of the common mode signal and the
DC component of the differential signal in the first stage of
the three op-amp DC IA are obtained by two feedback
loops, shown in Figure 1.
The first feedback loop returns the IA output signal via
the integrator K1 and controlled voltage sources V1 and V2.
The transfer function for the differential signal is given by
Vo ( s )
AD 0 s
,
(1)
=
A=
D (s)
Vd ( s ) ( s + 2 K1 AD 0 )
where AD0 is the band pass gain of the designed AC
amplifier,
 2R  R
(2)
AD 0= 1 + 2  4 ,
R1  R3

and a lower cut-off frequency is given by
K
(3)
f c = 1 AD 0 .
π
8
JORGOVANOVIĆ N, ET AL. AN IMPROVED AC-AMPLIFIER FOR ELECTROPHYSIOLOGY
The second feedback loop returns the extracted common
mode signal from the output of the first stage of the IA VC1
via amplifier K2 and the controlled voltage source V3. If
the value of the voltage V3 is given by

R 
(4)
V3 Vcm 1 + 1  ,
=
R2 
2

then the common mode voltage signal is completely
removed from the outputs of the first stage. In the
proposed circuit voltage source, V3 is controlled by the
second feedback loop (with amplifier K2). Therefore, the
expression for common mode suppression in the first stage
is
Vc1
R1 + 2 R2
1
(5)
=
≈
for R2 >> R1 .
Vcm R1 + 2 K 2 R2 K 2
Figure 1: Block diagram of the designed IA
As shown, the voltage sources V1 and V2 in the IA input
loop do not need to be "truly" floating. Instead of true
floating sources, it is satisfactory to provide two controlled
voltage sources referenced to the amplifiers with voltages
V3+V1 and V3-V2; this can be accomplished by using
ordinary op-amps, which will be described later.
U6 (R11, R13 and R15) (voltage V3-V2). The second
feedback loop returns the common mode signal (Vc1) from
the output of the first stage of the IA, formed at the middle
point of the series connection of the two equal resistors R5
and R6. The Vc1 signal is buffered by op-amp U8 and
amplified by an inverting amplifier based on op-amp U9,
resistors R25 and R26 and compensation capacitor C2
(voltage -V3). Gain of the inverting amplifier is set to 60
dB at DC. Due to frequency compensation, the low pass
cut-off frequency of the inverting amplifier is 15.9 Hz.
Therefore, amplification of the inverting amplifier is
reduced to 49.8 dB at 50 Hz. The output voltages of
amplifiers U6 and U7 are fed into the IA input loop via the
resistive network R17 - R18 - R19. This ensures attenuation
of the voltage difference V1-V2 by 53.5 dB and no
attenuation of voltage V3. The resistive network R17 - R18 R19 affects the low cutoff frequency through the constant
K1 from expression (3). The constant K1 also depends on
the integrator’s constant and the voltage divider R20-R21
(6):
R19
R21
1
.
(6)
K1 =
R20 + R21 R22 C1 R17 + R18 + R19
For the values of the components shown in Figure 2, the
constant K1 is equal to 6.49·10-6, which ensures a low
cutoff frequency of 0.01 Hz. The maximum DC signal
value that can be suppressed is limited by the maximum
output voltage of the integrator Vimax (defined by the power
supply voltage and op-amp output characteristics) and the
attenuation of the resistive network R17 - R18 - R19.
Additionally, Vimax is transformed into a voltage difference
between the outputs of the two adders, attenuated through
the resistive networks and subtracted from the input
differential voltage. Maximum DC suppression can be
described as
R19
.
(7)
Vin ( DC ) = Vi max
R17 + R18 + R19
For the components shown in Figure 2, with a power
supply of ±12 V, DC component suppression is 25 mV,
B. Circuit Design of the AC Amplifier
For testing purposes, an AC amplifier, shown in Figure which is sufficient for the recordings. The suppression can
2, was designed. The characteristics of the designed be increased with lower attenuation in the resistive
amplifier are the following: 1) a differential gain of 79.2 network R17 - R18 - R19.
dB and 2) a low cut-off frequency (0.01 Hz) and common As we have already stated, the resistive network R17 - R18 mode attenuation at the first stage of the IA (49.8 dB at 50 R19 ensures attenuation of the differential signal, but should
Hz). To achieve a CMRR that is as high as possible, a gain have no influence on the common mode. Expression (8)
of 59.2 dB is concentrated at the first stage. Therefore, the shows a relationship between the voltage on resistor R19
gain of the first stage and the gain-bandwidth product of (V19) and the common mode voltage Vc1:
R
R
the op-amps define the upper cut-off frequency. The
(8)
=
V19 K 2 K DS ( 11 − 12 )Vc1 ,
design was based on and tested with a frequently used opR15 R16
amp (OP27) so that an upper cut-off frequency of 8.8 kHz where K is the amplification of the common mode voltage
2
is achieved. The higher cut-off frequency, if required, V by amplifier U ,
c1
9
could be achieved with higher gain-bandwidth op-amps.
R25
1
.
(9)
K2 =
A standard three op-amp IA is based on op-amps U1, U2
R26 sC2 R25 + 1
and U3 and resistors R1, R2, R3, R4, R7, R8, R9, R10 and R19.
Feedback signals, linear combinations of the voltages KDS is the attenuation of the resistive network R17 - R18 V3+V1 and V3-V2, are defined by two adders: U6 (R11, R13 R19:
and R15) and U7 (R12, R14 and R16). The first feedback loop
R19
.
(10)
K DS =
returns output voltages from the IA via two branches: 1)
R17 + R18 + R19
attenuator R20-R21; integrator U4-R22-C1; and adder U7 (R12,
R14 and R16) (voltage V3+V1) and 2) attenuator R20-R21;
If the ratio R11/R15 is equal to R12/R16, the voltage on R19
integrator U4-R22-C1; inverter U5 (R23 and R24); and adder
is zero and the common mode signal will not be
JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL 19, 2009.
Figure 2: Circuit diagram of designed AC amplifier
transformed into a differential signal. To achieve equality
between R11/R15 and R12/R16, we used trimmer P1 that has
an influence on CMRR only. A mismatch in the resistor
values in the differential amplifier (U3, R7, R8, R9 and R10)
has moderate influence on CMRR because common mode
signals at the input are already attenuated by 49.8 dB at 50
Hz.
C. Simulation of the New AC Amplifier
The circuit’s stability and frequency analysis was
performed using Matlab and Simulink. According to the
schematic diagram in Figure 2, two Simulink models were
formed. One model describes the transfer function for the
differential mode signal of the whole amplifier (DM
model) and the other describes the transfer function for the
common mode signal for the first stage of the IA (CM
model). Bode frequency characteristics of the differential
gain are presented in Figure 3.
9
The differential gain is 79.2 dB, with a low cut-off
frequency of 10 mHz and a high cut-off frequency of 8.8
kHz. The low cutoff frequency is the result of an integrator
present in the feedback loop and does not depend on the
op-amp's characteristics. Moreover, the high cutoff
frequency depends on the op-amp's band-pass
characteristics. This can only be increased by using an opamp with a wider band-pass or by decreasing the gain at
the first stage of the IA.
Figure 4: Bode frequency characteristics of common-mode attenuation in
the first stage of the amplifier
Figure 3: Bode frequency characteristics of differential gain
Figure 4 shows the Bode frequency characteristics of the
common mode gain. The CM model describes the common
mode attenuation from the amplifier's input to output for
the first stage of the IA. The common mode signal
attenuation at the output of the first stage is 49.2 dB at 50
Hz. The open loop gain analysis shows that the system is
stable with a gain margin of 49.8 dB at 3.39 MHz; the
phase margin is infinity.
10
JORGOVANOVIĆ N, ET AL. AN IMPROVED AC-AMPLIFIER FOR ELECTROPHYSIOLOGY
III. TESTING OF THE AC AMPLIFIER
To test the new topology with two feedback loops, we
built a prototype without optimizing the power
consumption, the size of the prototype’s PCB area and the
circuit complexity. The reasons for skipping optimization
are the plans for integration of the amplifier into the single
chip.
Two types of prototype verification were performed: 1)
verification of the amplifier’s characteristics based on test
signals and 2) clinical verification by EMG recording in a
noisy environment. Data acquisition included a Tektronix
TDS3012 digital scope with a Tektronix differential
voltage probe ADA400A. A test signal was generated by
an Agilent 33220A signal generator.
A DC signal suppression test was performed by using the
sine wave input differential signal with an amplitude of 0.4
mVp-p, a frequency of 1 kHz and a DC offset of 0.8 mV.
The common mode signal on the input was set to zero in
this test. The result of the test is shown in Figure 5. The
upper waveform (Ch1) shows the differential signal at the
input of the amplifier. The bottom waveform (Ch2)
represents the output of the amplifier with a sine wave
signal with an amplitude of 4 Vp-p, a frequency of 1 kHz
and a DC offset equal to zero. Test results confirm
calculations and simulations and they were similar to the
other amplifiers with active DC signal suppression.
Testing of the common mode signal reduction in the
first stage of the amplifier was performed with a high
amplitude common mode input signal. We used a sine
wave test signal with an amplitude of 15.8 Vp-p and a
frequency of 50 Hz; the differential mode signal at the
input was set to zero. The results of the CMRR test are
presented in Figure 6a. The upper waveform (Ch1) shows
the common mode signal at the input of the amplifier. The
bottom waveform (Ch2) represents the common mode
signal at the output of the first stage of the amplifier.
According to the measured results, the common mode
signal attenuation in the first stage of the amplifier is 49.8
dB at 50 Hz. This result is a significant improvement
compared to the other amplifiers that had no common
mode reduction in the first stage.
To confirm the high total CMRR of the amplifier, the
output signal of the amplifier was recorded and the results
are shown in Figure 6b. The upper waveform (Ch1) shows
the common mode signal at the input of the amplifier. The
bottom waveform (Ch2) represents the common mode
signal at the output of the amplifier. According to the
measured results, the CMRR is 130 dB at 50 Hz.
Figure 5: Verification of the differential gain and DC-offset suppression.
Ch1 shows the input signal waveform to the amplifier. Ch2 shows the
output signal waveform.
The experimental verification of the amplifier on real
signals is demonstrated by the EMG recording in an
intentionally made, extremely noisy environment. The
tested amplifier was located very close to three switch
mode power supplies and GSM phones, also it was tested
without EMI filters, shielding, or a driven right leg circuit.
We recorded EMG signals by using two Neuroline
electrodes positioned over the Flexor Digitorum m.,
following the positioning instructions defined in the
SENIAM project [18]. The neutral (i.e., ground) electrode
over the Ulnar nerve was connected directly to the
common electrode of the amplifier. The recordings from
the contracted muscle of an able-bodied individual are
presented in Figure 7a with the time scale set to 10 ms per
division. Figure 7b shows three consecutive contractions
(weak, medium and strong) from the same muscle. The
time scale (horizontal axes) is 1 second per division.
Figure 6: Verification of the common-mode rejection in the first stage of the amplifier (left). Ch1: common-mode signal at the input of the amplifier;
Ch2: common-mode signal at the output of the first stage of the amplifier; Verification of the total common-mode rejection (right). Ch1: commonmode signal at the input of the amplifier; Ch2: common-mode signal at the output of the amplifier
JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL 19, 2009.
11
Figure 7 Short-time contraction of the Flexor Digitorum Muscle (left). Amplification of the amplifier is 79.2dB. Amplitude of the recorded signal is
2.2mVp-p; EMG signal of Flexor Digitorum Muscle (right). Amplification of the amplifier is 79.2dB. Amplitude of the low contraction is 0.6mVp-p,
amplitude of the medium contraction is 1mVp-p and amplitude of the high contraction is 2mVp-p
The upper waveform (Ch1) in Figure 8 shows the same
EMG signal with a different time scale setting (i.e., 400 ms
per division). The bottom waveform represents the
frequency spectrum of the signal.
As a result, almost the entire dynamic range of the
amplifier’s first stage is available for a useful component
of the input signal. The active DC offset suppression
allows DC coupling between the electrodes and the
amplifier, providing extremely high input impedance for
the designed amplifier. The entire design of the novel AC
amplifier, with simulations and test results, were presented
in this paper. A high CMRR of 130 dB at 50 Hz, a
differential gain of 79.2 dB, a cutoff frequency at 0.01 Hz,
a bandwidth of 8.8 KHz and, in particular, a common
mode reduction of 49.8 dB in the first stage of the
amplifier were achieved. Major characteristics of the
amplifier were confirmed by the test results. With these
characteristics, this amplifier can be used to record EMG
and other electrophysiological signals. The described
circuit is now under clinical testing as part of our virtual
EMG system [19] and [20].
REFERENCES
[1]
Figure 8: EMG signal of the Flexor Digitorum Muscle and its frequency
spectrum
The tests verify that the realized amplifier matches the
calculations and simulation results.
IV. CONCLUSION
The development of a new amplifier’s topology that has
characteristics
necessary
for
electrophysiological
recordings in a noisy environment was presented in this
paper. The novel topology is the active suppression of both
the DC component of the differential signal and the
common mode signal in the first stage of the three op-amp
IA. Active suppression of DC-offset was obtained by
feeding back an integrated form of the amplifier's output
that is equal to the input DC-offset, however with the
opposite sign. The active common mode signal
suppression at the first stage of the IA was obtained by
feeding back an extracted common mode signal from the
output of the first stage and presents an original
improvement in the electrophysiological amplifier’s
design.
B. B. Winter and J. G. Webster, "Reduction of interference due to
common mode voltage in biopotential amplifiers," IEEE Trans
Biomed Eng, vol. BME-30(1), pp. 58-62, 1983.
[2] N. V. Thakor and J. G. Webster, "Ground-free ECG recording with
two electrodes, " IEEE Trans Biomed Eng, vol. BME-27(12), pp.
699-704, 1980.
[3] J. C. Hunta and J. G. Webster, ”60-Hz interference in
electrocardiography,” IEEE Trans Biomed Eng, vol. BME-20,
pp.91-101, Mar. 1973.
[4] A. F. Pacela, ”Collecting the body’s signals,” Electronics, 40, (14),
pp. 103-112, 1967.
[5] R. Pallas-Areny and J. G. Webster,”Common Mode Rejectrion
Ratio in Differenctial Amplifiers,” IEEE Trans Biomed Eng, vol.
BME-40, pp. 669-676, 1991.
[6] R. Pallas-Areny, J. Colominas and J. Rosell, "An improved buffer
for bioelectric signals," IEEE Trans Biomed Eng, vol. BME-36(4),
pp. 490-493, 1989.
[7] R. Pallàs-Areny and J. G. Webster, "Analog Signal Processing,"
New York: Wiley, 1999.
[8] M. J. Burke and D. T. Gleeson, "A micropower dry-electrode ECG
preamplifier," IEEE Trans Biomed Eng, Vol. BME-47(2), pp. 155162, 2000.
[9] R. Pallas-Areny and J. G. Webster, " AC Instrumentation Amplifier
for Bioimpedance Measurements," IEEE Trans Biomed Eng, vol.
BME-40(8), pp. 830-833, 1993.
[10] Z. Nikolic, D. B. Popovic, R. B. Stein and Z. Kenwell Z.
"Instrumentation for ENG and EMG recordings in FES systems,"
IEEE Trans Biomed Eng, vol. BME-41. pp. 703-706, 1994.
[11] Yue-Der Lin, Chang-Da Tsai, Hui-Hsun Huang, Dah-Chuan Chiou
and Chien-Ping Wu, ”Preamplifier with a Second-Order High-Pass
12
[12]
[13]
[14]
[15]
JORGOVANOVIĆ N, ET AL. AN IMPROVED AC-AMPLIFIER FOR ELECTROPHYSIOLOGY
Filtering Characteristic,” IEEE Trans Biomed. Eng, vol. BME-46,
pp. 609-612, 1999.
H. W. Smit, K. Verton and C. A. Grimbergen, "A low-cost
multichannel preamplifier for physiological signals," IEEE Trans
Biomed Eng, vol. BME-34(4), pp. 307-310, 1987.
E. M. Spinelli and M. A. Mayosky, "AC coupled three op-amp
biopotential with active DC suppression," IEEE Trans Biomed
Eng, vol. BME-47(12), pp. 1616-1619, 2000.
E. M. Spinelli, N. Martinez, M. A. Mayosky and R. Pallas-Areny,
"A Novel Fully Differential Biopotential Amplifier With DC
Suppression," IEEE Trans Biomed Eng, vol.51, No.8, pp. 14441448, August 2004.
T. Degen and H. Jäckel, "A Pseudodifferential Amplifier for
Bioelectric Events With DC-Offset Compensation Using TwoWired Amplifying Electrodes," IEEE Trans Biomed Eng, vol.53,
No.2, pp. 300-310, February 2006.
[16] B.Gosselin, M. Sawan and C.A. Chapman, "A Low-Power
Integrated Bioamplifier With Active Low-Frequency Suppression,"
IEEE Trans Biomedical Circuits and Systems, vol.1, No.3, pp.
184-192, September 2007.
[17] E. M. Spinelli and M. A. Mayosky, “Two-electrode Biopotential
Measurements: Power Line Interference Analysis,” IEEE Trans
Biomedical Enginering, vol. 52, No. 8, pp. 1436–1442, Aug. 2005.
[18] H. J. Hermens, B. Fredriks, R. Merlett, D. Stegeman, J. Block, G.
Rau, et al., SENIAM European Recommendations for Surface
electromyography, Roessingh Research and Development BV,
1999.
[19] N. Stojanova, M. Petrovic and D. B. Popovic, "New approach to
laboratory measurements – virtual instruments," J Applied
Measurem, vol. 12, pp. 17-24, 1999.
[20] M. Piperski, D. B. Popović, “Virtual instrument for measuring
evoked potentials of peripheral nerves: electroneurography and
EMG reflexology”, in Proc. 8th Information Technology Conf,
Žabljak, February 22-26 2004, pp. 21-24.