J. Biomedical Science and Engineering, 2009, 2, 521-525
doi: 10.4236/jbise.2009.27075 Published Online November 2009 (http://www.SciRP.org/journal/jbise/
JBiSE
).
Published Online November 2009 in SciRes. http://www.scirp.org/journal/jbise
An oversampling system for ECG acquisition
Yu Zhou
School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai, China.
Received 31 July 2009; revised 31 August 2009; accepted 3 September 2009.
ABSTRACT
Traditional ECG acquisition system lacks for flexibil-
ity. To improve the flexibility of ECG acquisition sys-
tem and the signal-to-noise ratio of ECG, a new ECG
acquisition system was designed based on DAQ card
and Labview and oversampling was implemented in
Labview. And analog signal conditioning circuit was
improved on. The result indicated that the system
could detect ECG signal accurately with high sig-
nal-to-noise ratio and the signal processing methods
could be adjusted easily. So the new system can sat-
isfy many kinds of ECG acquisition. It is a flexible
experiment platform for exploring new ECG acquisi-
tion methods.
Keywords: ECG Acquisition; Oversampling; DAQ; Lab-
view
1. INTRODUCTION
ECG acquisition has been a mature technology since
Doctor Willam Enthoven in Holland detected human
ECG firstly. ECG can reflect the physiology status and
pathology status of heart. Heart diseases can be detected
in time by ECG. So ECG is used widely in clinic [1].
ECG can be divided into regular ECG, body surface po-
tential mapping, high frequency ECG, ventricular late
potentials electrogram, surface Hisbundle electrogram
and etc. Their detection objects are all with low ampli-
tude but distributing in different low frequency ranges
[2]. Traditional ECG acquisition systems are designed
according to different detection objects. Their analog
signal conditioning circuit including multistage amplifi-
ers and filters focus on special object. The disadvantage
is that the special ECG acquisition system only can de-
tect special object. So it lacks for flexibility and wastes
resources.
The DAQ card and Labview produced by NI Com-
pany provide a flexible solution for data acquisition and
processing [3,4]. DAQ card can detect signal accurately
with high speed. And digital signal processing can be
implemented easily in Labview. There are many works
using DAQ card and Labview in ECG. But they almost
focus on the ECG’s analysis and processing [5,6].
To resolve the problem about flexibility, analog signal
conditioning circuit is improved on and a new ECG ac-
quisition system is developed with DAQ card and Lab-
view.
Oversampling technology can exert the advantage of
high speed A/D and improve the signal-to-noise ratio of
signal detected. There has been detailed discussion about
the use of oversampling in biopotential detection [1,7,8].
This technology is implemented in Labview and used in
the new ECG acquisition system to improve the signal-
to-noise ratio of ECG.
2. SYSTEM DESCRIPTION
The oversampling system for ECG acquisition shown in
Figure 1 is consisted of analog signal conditioning cir-
cuit, DAQ card and Labview.
Research supported by The Special Funds for Scien-
tific Research for Selecting and Training Outstanding
Young Teachers in Shanghai Universities (slg08029).
The analog signal conditioning circuit condition the
ECG signal detected by electrodes to an analog signal
with certain amplitude and frequency restriction. The
analog signal is converted to digital signal by DAQ card
with high speed sampling rate. And oversampling and
other processing methods for the digital signal are im-
plemented in Labview.
3. ANALOG SIGNAL CONDITIONING
CIRCUIT DESIGN
Analog signal conditioning circuit included the circuit
from the electrodes to A/D input of DAQ card. To fit for
the need of the oversampling system for ECG acquisi-
tion, traditional circuit was improved on.
Research supported by The Special Funds for Scientific Research for
Selecting and Training Outstanding Young Teachers in Shanghai
Universities (slg08029).
Figure 1. Lock diagram of the oversampling system for
ECG acquisition.

Y. Zhou / J. Biomedical Science and Engineering 2 (2009) 521-525 522
3.1. Circuit Structure
The circuit was shown in Figure 2. As traditional circuit,
it included protection circuit, amplifying circuit based on
AD620, high-pass circuit for eliminating polarization
voltage, main amplifier, low-pass circuit and driven-
right-leg circuit. But the parameters set of analog filter
was different. In traditional circuit the filter parameters
were set according to the object detected and couldn’t be
adjusted. Thus it lacked for flexibility. In oversampling
system for ECG acquisition, the analog filter parameters
were set with 0.03Hz–4KHz. Although the filter para-
meters were also fixed, the frequency range was wide
and special digital filter parameters could be set further
in Labview. This scheme guaranteed the currency of the
system.
3.2. Isolated Circuit and Power
DAQ card and Labview worked based on PC. Isolated
circuit was needed to protect the object detected from
electric shock. HCNR200 was an optical coupler with
high linearity and fit for the oversampling system for
ECG acquisition. The isolated circuit based on HCNR
200 was shown in Figure 3. Isolated power supply was
from 9V battery. LM2576 and ICL7660 got energy from
9V battery and produced ±5V. Non-isolated power sup-
ply was from DC supply.
RG2
RG1
26
74 81
3
5
-
+
U1
AD620
R23 1.8k
C26 68p
C184.7u
R15 1k
R18
220k
R1 220K
R19
24.9k
R25
1k
R3 15K
D5
1N4148
R13
12.7k
C21 0.02u
C19
220p R12
1M
R22
1.8k
D1
P6KE39CA
R14
24.9k
RG1
RG2
FloatPosPower
FloatPosPowerFloatNegPower
FloatNegPower
Float NegPower
Float PosPower
Electrode+
Electrode-
Driving
FloatNegPower
FloatPosPower
FloatPosPower
Lift FloatNegPower
FloatNegPower
Lif t
FloatPosPower FloatNegPower
Float PosPower
R29
0
in+
D6
1N4148
in-
3
21
84
-
+
U28A
TL062
R24
586
5
67
8 4
-
+
U29B
TL062
C22
0.02u
C4
100p
C3
10p D3
3.6V
R2 220K
R5 15K
C2
100p D4
1N4148
5
67
84
-
+
U25B
TL062
D2
3.6V
R11
4.12k
D7
1N4148
C5
100p
R20 866k
3
21
8 4
-
+
U30A
TL062
R6 15K
C1
100p
R16 10k
R4 15K
Signal
AD620OUT
Figure 2. Analog signal conditioning circuit.
R26
51k
LC
1
LA
2
PD1C
3
PD1A
4PD2A 5
PD2C 6
NC1 7
NC2 8
k1 k2
U15
HCNR200
3
21
84
-
+
U27A
TL062
3
21
8 4
-
+
U29A
TL062
FloatPosPower
FloatNegPower
RealPosPower
FloatPosPower
RealNegPower
C25 1n
R27 200
C24
1n
Signal_ISO
Signal
R28 51k
Figure 3. Isolated circuit.
SciRes
Copyright © 2009 JBiSE

Y. Zhou / J. Biomedical Science and Engineering 2 (2009) 521-525 523
4. LABVIEW PROGRAM DESIGN
Labview program was the main body of the oversam-
pling system for ECG acquisition. It included the set of
DAQ card, digital filter, oversampling and etc.
4.1. Setting DAQ Card
DAQ6221 was a data acquisition card with PCI interface.
Its highest sample rate for single channel was 250KHz.
In the oversampling system for ECG acquisition, the
sample rate was set with 250KHz and continuous sam-
pling mode was chosen. And the reference voltage for
A/D in DAQ card could be chosen. ±5V was set because
the analog signal conditioning circuit had amplified the
signal enough.
4.2. Signal Filters
Filters including low-pass filter and 50Hz notch filter
were needed in Labview. In the basis of analog filters
with wide range, special digital filters were used for spe-
cial object detected in Labview. It was the flexibility of
the oversampling system for ECG acquisition. 50Hz-
interference caused by power line was an important
problem in biopotential detection [9,10]. Notch filter
was the tool to eliminate the interference. In Labview
digital 50Hz notch filter with more orders could get
much better result than analog notch filter. At the same
time the delay caused by digital filter was very small
because of the high-speed cpu of PC.
In experiments, for low-pass filter, type was set with
Butterworth, order was set with 50 and cut-off frequency
was set with 90Hz. For notch filter, type was set with
Bessel, order was set with 50.
4.3. Oversampling
The criteria of using oversampling were that the original
sample rate was high enough and the sample rate after
oversampling could satisfy Nyquist sampling theorem.
The sample rate in DAQ card was set with 50KHz. So
the sample rate after oversampling was 50KHz/256≈ 195
Hz if oversampling coefficient was set with 256. It was
enough for ECG to satisfy Nyquist sampling theorem
[1,7]. But if the oversampling coefficient was increased,
the Nyquist sampling theorem couldn’t be satisfied. If
sampling rate was improved with 4 multiples, the resolu-
tion of ADC could be improved with 1 bit [1,11]. So the
resolution of DAQ card’ ADC was improved with 4 bits.
Continuous sampling mode was chosen for DAQ card.
So continuous 256 samples were sampled and accumu-
lated and output. When this cycle went on, oversampling
result was get. Oversampling coefficient and filter pa-
rameters were both decided by the object detected and
could be set in Labview according to need. In experi-
ments, 250KHz was also used for sample rate in DAQ
card sometimes. Then the oversampling coefficient sh-
ould be changed to 1024 and the sample rate after over-
sampling was 244Hz. It was also right for ECG.
5. EXPERIMENTS DESIGNS AND
RESULTS
Some experiments were used to validate the design.
5.1. Oversampling Experiment
A sine wave with 20Hz and 5mV Vpp similar to ECG
produced by signal generator was input to DAQ card.
The results were shown in Figure 4. Oversampling re-
sult was better than the direct sampling result apparently.
FFT was applied for the two waveforms in Figure 4
and the results were shown in Figure 5. The main com-
Figure 4. Direct sample result (upside, sampling rate:
50KHz) vs. Oversampling result (downside, sample rate:
250KHz, Oversampling coefficient: 1024).
Figure 5. FFT for direct sample result (upside) vs. FFT for
Oversampling result (downside).
SciRes
Copyright © 2009 JBiSE

Y. Zhou / J. Biomedical Science and Engineering 2 (2009) 521-525 524
ponents in both FFT were in 20Hz. But there were many
components in other frequencies in the FFT for direct
sample result apparently. The FFT for oversampling re-
sult was clearer. Through careful observation, the value
in 20Hz in FFT for direct sample result was 316 and the
value in 50Hz which was the frequency of power line
interference in FFT for direct sample result was 1.46.
The corresponding values in FFT for oversampling re-
sult were 2454 and 10.3 respectively. Only considering
these components, the signal to noise ratio were 216 for
direct sample result and 238 for oversampling result. So
the FFT for oversampling result was clearer and its sig-
nal to noise ratio was higher. These were the reasons that
oversampling result was better than direct sample.
5.2. ECG Acquisition Experiment with 3
Electrodes
In the oversampling system for ECG acquisition, acqui-
sition mode with 3 electrodes could be chosen where
driven-right-leg electrode was included. Two electrodes
for detection were pasted in the left wrist and right wrist
respectively. The detection results were shown in Figure
6. The 1st was direct sampling signal sampled by DAQ
card, the 2nd was signal after oversampling, the 3rd was
signal after low-pass filter and the 4th was signal after
low-pass filter and notch filter from upside to downside.
It could be observed that the effect of the digital filters
was good and oversampling could improve the sig-
nal-to-noise ratio of ECG. The components of ECG were
shown clearly.
Figure 6. Sampling results with driven-right-leg circuit.
Figure 7. Sampling results without driven-right-leg circuit.
5.3. ECG Acquisition Experiment with 2
Electrodes
Also acquisition mode with 2 electrodes could be chosen
for the oversampling system for ECG acquisition where
driven-right-leg electrode was not included. Two elec-
trodes were pasted in the left wrist and right wrist re-
spectively. The detection results were shown in Figure 7.
The 1st was direct sampling signal sampled by DAQ card,
the 2nd was signal after oversampling, the 3rd was signal
after low-pass filter and the 4th was signal after low-pass
filter and notch filter from upside to downside. It could
be observed that the power line interference was worse
due to lack of driven-right-leg circuit before notch filter.
But after notch filter ECG became very clear and the
effect could be equivalent with the effect with driven-
right-leg circuit. So profiting from digital filter with high
orders, the detection result from 2 electrodes was satis-
fying. And oversampling could improve the signal-to-
noise ratio of ECG further.
5.4. ECG Acquisition Experiment in Left Arm
In ECG monitoring field, it was a problem waiting to
be resolved that how to acquire ECG with 2 electrodes
expediently. The correlative experiment was attempted.
An electrode was pasted in left wrist and the other was
pasted in the left shoulder. The results were shown in Fig-
ure 8 . It was observed that the ECG could be detected from
one arm and oversampling made result a little clearer.
SciRes
Copyright © 2009 JBiSE

Y. Zhou / J. Biomedical Science and Engineering 2 (2009) 521-525
SciRes Copyright © 2009
525
of Bioelectricity Signal Detection Based on Oversam-
pling, Acta Electronica Sinica, 36, 1465–1467.
[2] Wang, B. H., (2003) Biomedical Measurement and In-
strument, 1st Edition, Fudan Press, Shanghai.
[3] Koutroulis, E. and Kalaitzakis, K., (2003) Development
of an integrated data-acquisition system for renewable
energy sources systems monitoring. Renewable Energy,
28, 139–152.
[4] Yeung, K. and Huang, J., (2003) Development of a re-
mote-access laboratory: a dc motor control experiment.
Computers in Industry, 52, 305–311.
[5] Wei, Y. C., Lee, Y. H., and Young, M. S., (2008) A Port-
able ECG Signal Monitor and Analyzer. The 2nd Interna-
tional Conference on Bioinformatics and Biomedical En-
gineer- ing. Shanghai, China, pp.1336–1338.
[6] Yeon, N., Sung, P., Kyu, H., et al., (2006) A Study of
Significant data Classification between EDR extracted
and frequency analysis of Heart Rate Variability from
ECG using Conductive textile. World Congress on
Medical Physics and Biomedical Engineering. Seoul,
Korea, pp.4100–4103.
Figure 8. Sampling results from one arm (direct sample re-
sult (upside), Oversampling result (middle), result after over-
sampling and low-pass filter and notch filter (downside)).
6. CONCLUSIONS
[7] Li, G
., Zhang, L. J., Lin, L., and He, F. (2008) Weak Sig-
nal Detection Based on Over-Sampling and Saw-Tooth
Shaped Function. ACTA ELECTRONICA SINICA, 36,
756–759.
The results of experiments indicate that the new over-
sampling system for ECG acquisition could detect ECG
accurately, digital signal processing can be implemented
easily in the system with DAQ card and Labview, ECG
can be detected clearly with both 3 electrodes mode and
2 electrodes mode profiting from digital filters with high
orders, oversampling improve the signal-to-noise ratio of
ECG especially in one arm ECG detection. And the new
system is flexible because the signal processing methods
can be adjusted easily in Labview such as filter parame-
ters and oversampling coefficient. So the oversampling
system for ECG acquisition with flexible set and high
signal-to-noise ratio can satisfy many kinds of ECG de-
tection.
[8] Goes, J., Paulino, N., Pinto, H., Monteiro, R., Vaz, B.,
and Garcao, A. S., (2005) Low-power low-voltage CMOS
A/D sigma-delta modulator for bio-potential signals
driven by a single-phase scheme. Circuits and Systems I:
Regular Papers, IEEE Transactions on, 52, 2595– 2604.
[9] Huhta, J. C. and Webster, J. G. (1973) 60-Hz Interference
in Electrocardiography. IEEE Transactions on Biomedi-
cal Engineering, 20, 91–101.
[10] Spinelli, E. M. and Mayosky, M. A., (2005) Two-electrode
biopotential measurements: power line interference ana-
lysis. Biomedical Engineering, IEEE Transactions on, 52,
1436–1442.
[11] Oppenheim, A. V., Schafer, R. W., and Buck, J. R., (1999)
Discrete-time Signal Processing, 2nd Edition, Prentice
Hall, New Jersey.
REFERENCES
[1] Li, G., Zhang, L. J., Lin, L., and He, F., (2008) Research
JBiSE