Proceedings of 20121CME

International Conference on Complex Medical Engineering

July I - 4, Kobe, Japan

Development of a Capacitive Coupling Electrode for

Bioelectrical Signal Measurements and Assistive Device
Use
Alexsandr Igorevitch Ianov, Hiroaki Kawamoto and Yoshiyuki Sankai
Graduate School of Systems and Information Engineering
University ofTsukuba
Tsukub� Ibaraki305-8573, Japan

{ianov, kawamoto, sankai}@golem.kz.tsukuba.ac.jp

Abstract - The human body is known to produce several • EMG measurements are used to interface with
different electrical patterns and signals derived from bioelectrical exoskeletons such as the HAL robot suit[3-6] or prosthetic
activity. Bioelectrical signals such as the cardioelectrical and
limbs[7 -8].
myoelectrical signals have several medical applications. In order
Wet disposable electrodes such as the Vitrode
to acquire those signals, low impedance contact electrodes are
(Nihonkohden, Japan) capable of sticking to the surface of the
commonly used. However, because these sensors require skin
preparation or application of conductive gels, placing these
skin are used for these measurements. Because these
sensors can be difficult, time consuming and uncomfortable task electrodes are passive electrodes, i.e., do not have integrated
for the patient. Moreover the presence of body hair and sweat can electrical amplification, on a dry state they are very susceptible
be a source of noise and cause signal depletion. This paper to noise and may not be able register robust bioelectrical
proposes a noncontact capacitively coupled electrode based on signals on the skin. Even while wet, due to the relatively high
high impedance input, required to detect weak electrical fields. output impedance of the electrode, the signal can be distorted
Noncontact capacitively coupled electrodes rely on reacting to
by cable movements or contact of the electrode with other
electrical field variations caused by bioelectrical activity,
solid surfaces. The saline(Ag/AgCI) gel that is used between
therefore eliminating the need to maintain resistive contact
the carbon electrode lead and the skin of the patient reduces
between the skin and the electrode. The sensor high impedance
input of 1 teraohms allowed only 0.3 to 1 picoampere input
the electrode input impedance thus, thus increasing the
current to flow between the sensor plate and the preamplifier conductivity and the signal noise ratio. Furthermore, these wet
unit. Experiments focused on the recording of electromyogram electrodes are not capable of staying still on a fixed spot on the
and electrocardiograms over clothing. Data was collected and skin using the conductive gel alone, especially when the
compared with data from resistive electrodes. We confirmed that patient is moving. In order to firmly attach the adhesive tape is
our electrodes are capable of collecting bioelectrical data that can placed around and over the area above the electrode on the
be used for assistive devices interfacing without direct contact
skin. However the use of gel and adhesive in wet electrodes
with the skin of the subject. Such electrodes can replace currently
have three major drawbacks[9].
used resistive contact electrodes improving the reliability of
collected signals and increasing user-friendliness of systems that
require daily, long term bioelectrical signal monitoring.

Index Terms - electrode, myoelectrical signals, capacitive

sensing.

I. INTRODUCTION

Cardioelectrical and myoelectrical signals generated by

muscle activity, such as the electrocardiogram(ECG) and
electromyogram(EMG), have become an important source of
information about the human body. As shown in Fig.I, such
signals have become important in a wide selection of fields,
ranging from medicine and rehabilitation to sports and
entertainment. Below are a few examples on the medical field
we would like to focus:
• ECG measurements done by a Holter Monitor for long
periods of time are frequently used to diagnose cardiac
diseases or measure cardiac stress under physical
exercises[I-2].
Fig. 1 Several Proposed Uses for Bioelectrical Signals

978-1-4673-1618-7112/$31.00 ©2012 IEEE 593

1. Both gel and adhesive may cause skin irritation and II. BIOELECTRICAL SENSING THROUGH CAPACITI VE COUPLING
allergic reactions.
Capacitive sensing is used to measure bioelectrical signals
2. Depending on the measurement type, the conductive gel
by performing AC coupling between the electrode and the skin
may dry degrading the signal and the adhesive may wear
of the patient. The equivalent circuit of a capacitive sensing
off due to sweat and muscular contractions.
system is given in Fig.2. By approaching a metal plate
3. Body hair removal is required on the areas bellow the
connected to a electronic circuit to the skin, a condenser
electrodes.
between the metal plate and the skin is formed. The skin­
Dry electrodes have been developed[lO] in order to
electrode-amplifier system can be modeled as a RC based high
eliminate the need to use of gel and adhesive. However, in
pass filter connected to ideal op-amp buffer. The capacitance
some cases, skin preparation such as body hair removal and
of the skin-metal plate capacitor is given by
cleaning. Plus constant electromechanical skin contact is still a
requirement. Being able to quickly attach or remove required £TIR2

C
(1)
sensors is an important step towards the adoption of assistive = -d-
devices and exoskeletons in daily life. assuming we use a circular metal plate with radius R from a
In order to increase the usability of the electrodes, distance d from the skin. E is the dielectric constant. Because
noncontact electrodes based on capacitive coupling have been there are various types of textiles, each with its own dielectric
proposed[ll]. Recently capacitive electrodes have gotten constant and because most of them are no continuous and are
precise enough to record ECG signals over the clothing[12]
filled with air, we can assume that E is the dielectric constant
and under bed sheets[13]. The major disadvantage of these
of the air. Setting distance d to around 1.0 [mm] and R to 15
electrodes is that due to the ultra-high impedance input they
[mm], the capacitance C of the skin-electrode capacitor is
are very susceptible to noise originated from electromagnetic
approximately 2.0[pF].
sources such as hum noise and electrostatic field noise from
In order to measure generated by common bioelectrical
nearby objects. In order to reduce this noise careful shielding
signals, which oscillate at frequencies of 0.1 [Hz] to 40[Hz]
and signal conditioning techniques are used. However, the use
for cardioelectrical signals and 20[Hz] to 500[Hz] for
of those techniques also increase the size of the electrode
myoelectrical signals, a high input impedance of at least 1.0[T
significantly, making them inappropriate for daily, multi
Q] is required.
channel recording required by the applications we mentioned
Common ground signal between the body and the circuit
above. Attempts to record EMG with capacitive sensors have
is realized by using a low impedance interface, such as a
also been made by embedding the electrode preamplifier
conventional dry or wet electrode. While it's possible to
circuit into the textile[14]. This approach would allow the
record ECG signals without a ground, especially when
creation of a sensing suit for capacitive EMG measurement but
measuring weaker EMG signals, better signal noise ratios are
it would still require the patient to change clothes because the
obtained by maintain a robust ground[10][12].
method was highly dependent on the textile material. Also by
embedding electrodes in the clothing, we must assure that each
III. MEASUREMENT SYSTEM
suit fits the wearer perfectly and we also lose sensor placement
freedom. A. Capacitive Coupling Electrode
In this paper we propose an electrode capable of recording In chapter II, we calculated that a preamplifier component
cardioelectrical and myoelectrical signals through clothing with a input impedance of at least 1.0[T Q ] is required.
using capacitive coupling configured with minimal impedance Considering the available components in the market, we chose
values for maximum noise resistance, thus allowing the to use an instrumentation amplifier with input impedance of
electrode to be small but precise enough for exoskeleton and 3.3[T Q].
assistive device use and long term daily monitoring 1--------------------------------
I
Electrode EquivalentCircuit
bioelectrical activity monitoring. In order to reduce the size of Human Body :
I
Equivalent Circuit I
I

i Skin-ElectrodeCapacitor
the electrode, we propose choosing a preamplifier component I

that has smaller but high enough impedance to be capable of

!=;
H>
sensing through capacitive coupling the small bioelectrical
voltage variations while still being strong against noise,
reducing the need for space consuming advanced shielding
techniques. Since wet electrodes such as the Vitrode have the Generator
(Muscle, Heart) To Target
best signal noise ratio, we compare data from our electrodes Device
against data from Vitrode electrodes and discuss how viable
they are for the applications we mentioned in the beginning of
-1_H+_____ _-I
- - - -
L _________________
L..
_ _ _______________________
this chapter. ///// Common Ground

Fig.2 Capacitive sensing system

594
In contrast, capacItIve coupling electrodes that appear in
previous researches use amplifiers with over 100[T Q ] IV. EXPERIMENTS AND RESULTS
[13][14], which are significantly weaker to electrostatic and
A. EMG Measurement Using Noncontact Electrode and Wet
hum noise. Furthermore, the used amplifier not only offered
Electrode
the required input impedance, but by being a high quality
When interfacing with assistive devices and exoskeletons,
instrumentation amplifier, the preamplifier could be
myoelectrical signals such as the EMG are used. The device
configured by using less components and with a simpler
EMG sensors rely on wet electrodes, so for the developed
printed circuit board pattern than previous capacitive
capacitive coupling electrode to be declared adequate for these
electrodes that use standard operational amplifiers. This
applications they must perform similarly to the common wet
contributes to smaller preamplifier board and less components
electrodes. In this experiment we investigate the correlation
that reducing potential sources of noise.
between the myoelectrical signals such as the EMG data
Based on the calculations done on Chapter II, we build
collected from a pair of noncontact electrodes with the EMG
our electrode using a 3[cm] diameter stainless steel plate.
data collected simultaneously for a pair of conventional
Stainless steel plates were chosen as electrode leads due to the
Vitrode disposable wet electrodes. Therefore we investigate
low bias current they generate when connected to the
the correlation of the data collected by both types of electrodes
preamplifier. The preamplifier board was allocated and
and verify the nature of the data collected by the developed
soldered with its input lead on one of the plate faces. The
capacitive coupling electrodes.
plate-board system was housed inside a acryl case. The case
Two developed noncontact electrodes are placed over the
had its interior covered with cooper tape which was also
right biceps of the experiment participant. Between the right
connected to the ground, creating a low profile shield. The
biceps skin and the electrode metallic plate a piece of cotton
total size of the electrode was of 3.2[cm] with a thickness of
with thickness of 1 [mm] is placed in order to simulate
8[mm]. Thickness is mostly due to the cable connector used.
clothing. Bellow the clothing, 2 Vitrode type D electrodes are
The assembled electrode is pictured in Fig.3 with the
placed as close as possible from the noncontact electrodes.
preamplifier board.
Ground signal is taken from the back of the right hand of the
The developed electrode also contained a circuits to
participant.
minimize the effects of input bias current of the
The experiment is performed by contracting the right
instrumentation amplifier.
biceps every 5 [s] and then relaxing for another 5 [s]. The
experiment settings block diagram is shown in Fig.5. The total
B. Data Recording System
measurement time is of 20 [s].
The developed electrode data recording and evaluation Developed
system consisted of 3 stages. The first stage being a second Electrodes

instrumentations amplifier that receives analog signals from

two capacitive electrodes and output the amplified difference
between them. A second stage is responsible for conditioning
the signal for the AD converter. The second stage is consisted
of a second order band pass filter and a voltage level
converter. And a final stage composed by a 16 bit AD
converted connected through a SPI channel to a
microcontroller. Sampling is performed at 1 [kHz]. Data was
transferred from the controller to a PC through a USB2.0 port.
The common ground is connected to the back of the hand Common Ground
/////
Vitrode type 0
of the user through a Vitrode type D disposable wet electrode.
A block diagram of the data recording system is shown at Fig.4 Data Recording System

3kgweight

Fig.5 Setting for Experiment A

595
 The recorded experiment data is shown in Fig.6. From the The results show that the developed noncontact electrode
results we can see that the EMG data collected from the is capable of recording signals with variable intensities and
Vitrode pair of electrodes overlap most of the data collected from the RMS values, we verify that the heavier the load the
by the noncontact electrodes. The calculated correlation stronger is the signal.
coefficient for this dataset was of 0.93, showing that the
developed capacitive electrodes are capable of collecting the C. ECG Measurements Over Clothing
signals originated from myoelectrical activity and that both the While using exoskeletons, EMG is used to interface and
noncontact electrode and vitrode disposable wet electrodes are actively communicate with the device. However, by reading
capable to collect very similar EMG signals. other bioelectric signals such as ECG, we can also evaluate or
estimate the amount of physical stress on the patient and adjust
B. Biceps EMG Measurements Under Variable Loads the exoskeleton control parameter accordingly, such as
When interfacing with exoskeletons it is important for the assistive gain. In this experiment we verify if our electrode is
patient to be able to control the output torque of the device. A capable of recording ECG signals with fidelity by verifying if
common method of setting the output is by measuring the the component waves of a standard ECG signal are present or
amplitude of the EMG signal used to drive the device[3-4]. not and by checking if the ECG signal is consistent over a long
Therefore the purpose of this experiment is to verify if the period of time.
developed noncontact electrode is capable of collecting EMG In this experiment two electrodes following standard ECG
data from a muscle at different loads and is adequate to be electrode placement methodes. The subject was wearing a
used as an interface sensor for exoskeletons and assistive cotton shirt with average thickness of 1 [mm] and the
devices. electrodes were placed above the shirt. Ground signal is taken
The experiment setting is very similar to the previous from the back of the right hand of the hand of the participant.
experiment. However in this case the participant keeps his Data was recorded for 20 [s] while the participant stayed seat
biceps under continuous stress for 10 [s]. Data is measured for comfortably on a chair. Recorded data is shown in Fig.8.
loads of 2[kg], 4[kg], 6[kg], 8[kg] and lO[kg]. Data is also From the results we can see that our electrode was capable
measured when there is no load. Similar to the previous of recording ECG signals consistently over a period of time of
experiment, a pair of noncontact electrodes is placed above 20[s]. Also by plotting the data from the first three seconds in
right biceps with a piece of cotton cloth with average thickness Fig.9, we are able to verify that the electrode recorded all the
of 1 [mm] between the electrode and the skin to simulate major component waves of the standard ECG waveform.
clothing. Ground is taken from the back of the right hand of
the participant. Acquired results are shown in Fig.6. The root
mean square(RMS) value for each case is calculated and
displayed inside its corresponding graph.

2.5

">
'-"
1.5

-1.5

Time (s)
Fig.6 Experiment A Data Set - Simultaneous Recording of Biceps EMG with capacitive coupling and wet electrodes

596
 ��
.,
DO,

!i
� ' r�--�------�--��----�--------�

a.,
!:l-'
c.

No weight Load:6kg

.,.l-+-
-- --+----+---R-'-'l-
-'- �
'--'-'S"__'_i:p
:=-. =- Y, ---+----+----.---1
1-" 7.r-= 1==2--f- RMS:0.5501 V
Time (5)

�� �
.,
DO,
.,
..
� ,I. .".r" ,.i<ill "U.. H .L, , , 1, ,, ", ','. . ,..I,L .j.,A, !i
'0
�. T . """" "'" ..,,' "" '''' f ''I' " ,. " 1' 1"'" ' ". >
�
::l-' !l
!l- !l
C.

0.2 a
Load:2kg
RMS:O.2589V
" .l-�--�--+---t--�,----t---+---��--I.
Time (5)

�.��--�--+-���,���+---��--�.
Time (5) Time (5)
Fig.7 EMG recorded by the capacitive coupling electrodes at varying loads

V. DISCUSSION

In this study we developed a capacitive coupling electrode

with about the same dimensions of a standard wet electrode,
but differently from the wet electrodes we did not have to
apply any conductive gel between the skin and the electrode
lead. Sweat and other long term problems responsible for the
degradation the bioelectrical signal of wet electrodes were not
"
Time (5) observed during any of the experiments. We did not have to
Fig.8 Consistent ECG recordings over the period 0[ 20 [s] perform any type of skin preparation nor remove clothes from
the experiment participant. Also differently from other
R capacitive coupling sensing techniques target at bioelectrical
signal measurement, we used a preamplifier unit with 100
times lower input impedance, strong against electrostatic
noise, with a minimal amount of components, minimal circuit
board size and shielding but that was capable of recording
signals very similarly to conventional wet electrode interfaces.
During experiment A in section IV there are two reasons
for the correlation coefficient between the data collected by
u
the wet electrodes and the data collected by the developed
Time (s) noncontact electrodes to be 0.93 instead of 1. One reason is
Fig.9 Zoomed in ECG

597
because the noncontact electrode and its correspondent wet In future works we plan to integrate our developed
electrode were not placed in the exact same place. Between the noncontact electrodes in to the robot suit HAL and evaluate
center of the noncontact electrode and the center of the wet the performance of the electrodes under real world conditions
electrode there was a distance of 3[cm]. Previous research as well as developing optimized versions of the electrode
have shown that 3[cm] is a distance big enough to produce based on the results of those experiments.
create distortions between the data collect at two points[15].
The second reason is, from formula (1), because the
REFERENCES
noncontact electrode output is also a function of the distance
[1] P. Bjerregaard, "Mean 24 hour heart rate, minimal heart rate and pauses in
between the electrode metal plate and the skin, any variation in
healthy subjects 40?79 years of age, "Eur Heart J (1983) 4 (I): 44-51.
this distance will reflect on the output of the electrode making [2] R. Wolfe, et aI, " Arrhythmias in patients with valvar aortic stenosis,
it susceptible to motion artifacts. As we can observe in Fig.6 valvar pulmonary stenosis, and ventricular septal defect. Results of 24-
there are spike in the noncontact electrode data set which are hour ECG monitoring , " Circulation. 1993 Feb;87(2 Suppl):189-101.
[3] S. Lee and Y. Sankai, "The Natural Frequency-Based Power Assist
no present in the wet electrode data set. And those spikes all
Control for Lower Body with HAL-3", Proc. of International
appear when the contraction starts or ends, hinting that the Conferenceon Systems, Man and Cybernetics (SMC2003), pp.1642-
actual muscle movement may be creating motion artifacts on 1647, 2003
the developed electrode. [4] T. Hayashi, H. Kawamoto and Y. Sankai, "Control Method of RobotSuit
Experiment B in section IV showed that our electrode is HAL working as Operator's Muscle using Biological and Dynamical
Information", Proc. of IEEE/RSJ International Conference onIntelligent
capable of recording different signals when the observed Robots and Systems (IROS 2005), pp.3455-3460, 2005
muscle is under different loads. This results shows that our [5] H. Kawamoto and Y. Sankai, "Power assist method based on
electrode has the potential of being used as the interfacing PhaseSequence and muscle force condition for HAL", Advanced
sensor for exoskeleton and other asslstlve devices. Robotics,voI.l9, no.7, pp.717-734, 2005
[6] K. Suzuki, G. Mito, H. Kawamoto, Y. Hasegawa and Y. Sankai,
Furthermore in experiment C in section IV, we verified that "Intention-Based Walking Support for Paraplegia Patients with RobotSuit
our developed electrode is capable of correctly recording HAL, Advanced Robotics", Vo1.21, No.I2, pp.144I-1469, 2007
ECG. This results show that the developed capacitive coupling [7] Saridis, George N.; Gootee, Thomas P.; , "EMG Pattern Analysis and
electrode has the potential to be used not only as an interfacing Classification for a Prosthetic Arm ," Biomedical Engineering, IEEE
Transactions on , voI.BME-29, no.6, pp.403-412, June 1982
tool but also as a diagnosing and evaluation tool for assistive
[8] Au, S.K.; Bonato, P.; Herr, H.; , "An EMG-position controlled system for
devices. However, all the experiments in this paper were an active ankle-foot prosthesis: an initial experimental study, "
conducted in ideal conditions, for a short period of time. In Rehabilitation Robotics, 2005. ICORR 2005. 9th International
order to verify how adequate are the developed electrodes, Conference on , vol., no., pp. 37 5- 379, 28 June-l July 2005
[9] A Searle and L Kirkup, " A direct comparison of wet, dry and insulating
long term experiments under real world conditions are still
bioelectric recording electrodes, " Physiol. Meas., 21 271, 2000
necessary. In particular, how the shield would perform near [10] Burke, MJ.; Gleeson, D.T.; , "A micropower dry-electrode ECG
high power motors present in exoskeletons and how the preamplifier, " Biomedical Engineering, IEEE Transactions on , vo1.47,
electrodes would perform when hit by external objects, in no.2, pp.155-162, Feb. 2000
[11] P.C.Richardson, "The insulated electrode: A pasteless
contact with high amounts sweat or when displaced by the
electrocardiographic technique " 20th Conf. on Engineeringin Medicine
muscular movements of the user. and Biology p 15.7
In this paper, during the experiments, we also proposed [12] CJ. Harland, T.O. Clark and RJ.Prance, "Electrical potential probes -
the use of a single wet electrode in order to create a robust new directions in the remote sensing of the human body, "Meas. Sci.
Technol., vol. 13 (2002), 163-169
ground between the user and the electronic system. Data from
[13] Y. Yama and A. Ueno, "Unrestrained Facile Measurement of Narrow­
all experiments have shown that our electrode performed very band ECG and Respiratory Variation in Infants with a Capacitive Sheet­
similarly to conventional electrodes, therefore eliminating the type Sensor, " Transactions of Japanese Society for Medical and
need to a robust ground connection using wet electrodes. For Biological Engineering. 47 No. I: 42-50, 2009
conventional electrodes, a ground connection between the user [14] T. Linz, L. Gourmelon and G. Langereis, "Contactless EMG sensors
embroidered onto textile, "4th International Workshop on Wearable and
can be formed by simply establishing contact between the user Implantable Body Sensor Networks (BSN 2007), Vo1.I3, pp.29-34, 2007
skin and a grounded metallic surface connected to the system, [IS] G Rau, C Disselhorst-Klug, "Principles of high-spatial-resolution
such as a case or frame. In order to maximize the usability of surface EMG (HSR-EMG): single motor unit detection and application in
the system, we can propose the same for the developed system the diagnosis of neuromuscular disorders, "Journal of Electromyography
and Kinesiology, Vol. 7, Issue 4, pp. 233-239, 1997
and design a wearable system that would guarantee a ground
connection as long as the user is performing recordings.

VI. CONCLUSION AND FUTURE WORKS

In this paper we proposed an electrode capable of

recording EMG and ECG through clothing through capacitive
coupling for exoskeleton and assistive device use and long
term daily monitoring bioelectrical activity monitoring for
medical applications.

598