ADDIS ABABA UNIVERSITY

ADDIS ABABA INSTITUTE OF TECHNOLOGY

DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING

Design and Simulation of Front End 3-in-1 EEG, ECG, EMG Bio-Potential Signal
Acquisition System

A thesis submitted to School of Graduate Studies and Research in Partial Fulfillment of

the Requirement for the Master’s Degree

By: Yonas Worede Legesse

June, 2020
Declaration

I, the undersigned, declare that this thesis work is my original work, has not been presented for a
degree in this or any other universities, and all sources of materials used for the thesis work have
been fully acknowledged.

Name: Yonas Worede Signature: _____________________

Place: Addis Ababa, Ethiopia

Date of Submission: __________________________

This thesis is submitted for examination with my approval as university advisor.

Advisor: Daniel Dilbie Signature: _______________________

 ADDIS ABABA UNIVERSITY

ADDIS ABABA INSTITUTE OF TECHNOLOGY

DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING

Design and Simulation of Front End 3-in-1 EEG, ECG, EMG Bio-Potential Signal
Acquisition System

By: Yonas Worede Legesse

APPROVED BY BOARD OF EXAMINERS

Chairman, Department of Graduate Committee Signature

Daniel Dilbie
Advisor Signature

Internal Examiner Signature

External Examiner Signature

 Acknowledgement
First and foremost, my sincerest appreciation goes to my advisor Daniel Dilbie, who has supported
me throughout my thesis work. His in-depth knowledge in practical design of electronic devices
and personal motivation has inspired me and pushed me in breaking new barriers and doing
detailed research. He has taken personal interest in my work through commentary and validation
and provided me with options when I was faced with tremendous challenges. Additionally, I would
like to express my gratitude to Nebeyou Yonas who has thoroughly evaluated the documentation
and constantly provided constructive feedback until completion without holding back.

Even though it is a platform maintained by thousands of participants, I would also like to express
my gratitude for engineers that participate in online forums with special emphasis to the online
portal of Texas Instruments. Their online blogs and webinars have helped me a lot when it comes
to familiarizing myself with new products and performance analysis talks.

Finally, I would like to thank friends and family who have provided me with emotional support
and always held me at a high standard. And a very special thanks goes to my mother whose support
is impossible to put into words. She has been constantly pushing me to be the best that I can be
and providing me with financial support when I was in need of it.

I
 Abstract
The last couple of years have given birth to meticulously mapped and innovative solutions in
regards to product as well as research of medical analysis tools and diagnostic equipment. The
industry has shown a major transformation on the general process of diagnosis tools providing
flexibility and enhanced accuracy. However, despite the astonishing progress of the bio-medical
industry, the status of medical provision is still a concern in third world countries. The
inaccessibility and unaffordability of medical equipment in such countries needs immediate
attention as many people fall prey to this problem which can be solved through the provision of a
supplementary solution that can aid the process of preliminary diagnosis.

In this thesis, the design of a front-end system for EEG, ECG and EMG signal acquisition is done.
The design addresses the problem of medical provision in under developed nations by providing a
supplementary hardware that is portable making it cost-efficient and readily available. Moreover,
it extends the research aspect in the area through the combination of a 3-in-1 signal acquisition
hardware and optimizing the design in regards to performance, complexity and scalability.

Owing to the fact that the signals operated by the hardware are very weak in nature, the utilization
of low noise amplifiers with very high common-mode rejection ratio and gain adjustment is
critical. Moreover, the implementation of analog-to-digital conversion needs a thorough analysis
in regards to the architecture, resolution and area of application. Accordingly, the design in this
thesis is specifically done so as to improve the performance in regards to noise cancellation,
minimization of filter circuitry, number of channels and overall circuit complexity.

Verification of the design is done with the co-simulation of PSPICE and SIMULINK. The
simulation is carried out for individual cases of EEG, EMG and ECG application by using
physiological signals of patients from PhysioNet.org through the addition of noise signals to mimic
actual physical application of the hardware. Furthermore, the output is analyzed and compared
with existing products and previous researches in the area which yielded a 21% improvement in
common-mode rejection ratio and a 33% increase in channel capacity.

II
Table of Contents
Acknowledgement ......................................................................................................................................... I
Abstract ......................................................................................................................................................... II
Table of Contents ......................................................................................................................................... III
List of Figures ............................................................................................................................................... VI
List of Acronyms ........................................................................................................................................... IX
Chapter 1....................................................................................................................................................... 1
Introduction .................................................................................................................................................. 1
1.1 Overview ............................................................................................................................................. 1
1.2 Theoretical background and design principles ................................................................................... 2
1.2.1 Signal characteristics .................................................................................................................... 2
1.2.2 Signal measurement type and standards .................................................................................... 5
1.2.3 Data acquisition system architecture .......................................................................................... 7
1.3 Problem statement ........................................................................................................................... 10
1.4 Objectives.......................................................................................................................................... 11
1.4.1 General objective ....................................................................................................................... 11
1.4.2 Specific objectives ...................................................................................................................... 11
1.5 Scope of the study ............................................................................................................................ 11
1.6 Significance of the study ................................................................................................................... 11
1.7 Thesis outline .................................................................................................................................... 12
Chapter 2..................................................................................................................................................... 13
Literature review......................................................................................................................................... 13
2.1 Bio-potential signals and measurement practices............................................................................ 13
2.2 Electroencephalogram ...................................................................................................................... 14
2.3 Electrocardiogram ............................................................................................................................. 14
2.4 Electromyogram ................................................................................................................................ 15
2.5 Related works ................................................................................................................................... 16
2.7 Summary ........................................................................................................................................... 18
Chapter 3..................................................................................................................................................... 19
Methodology............................................................................................................................................... 19
3.1 Design and implementation procedure ............................................................................................ 19
3.1.1 Preliminary ................................................................................................................................. 22
3.1.2 Literature review and analysis ................................................................................................... 22

III
 3.1.3 System design ............................................................................................................................ 23
3.1.4 Simulation .................................................................................................................................. 23
3.1.5 Hardware design ........................................................................................................................ 24
3.1.6 Analysis and evaluation ............................................................................................................. 24
3.1.7 Conclusion .................................................................................................................................. 24
3.2 Tools and signal resources ................................................................................................................ 24
Chapter 4..................................................................................................................................................... 26
Proposed circuit architecture and implementation.................................................................................... 26
4.1 Overview ........................................................................................................................................... 26
4.2 System design ................................................................................................................................... 26
4.2.1 Goals and Metrics ...................................................................................................................... 27
4.2.2 Bio-potential signal conditioning ............................................................................................... 28
4.2.3 Analog to digital conversion ...................................................................................................... 32
4.2.4 System design review................................................................................................................. 35
4.3 Simulation ......................................................................................................................................... 35
4.3.1 Bio-potential signal sources ....................................................................................................... 37
4.3.2 System simulation models ......................................................................................................... 39
4.4 Hardware Design ............................................................................................................................... 43
4.4.1 Overvoltage protection .............................................................................................................. 44
4.4.2 Signal amplification .................................................................................................................... 45
4.4.3 Analog to digital conversion ...................................................................................................... 46
4.4.4. PCB design ................................................................................................................................. 47
4.5 Simulation results and discussion ..................................................................................................... 50
4.5.1 Signal amplification .................................................................................................................... 50
4.5.2 Effects of noise ........................................................................................................................... 51
4.5.3 CMRR analysis ............................................................................................................................ 55
4.5.4 PSRR analysis .............................................................................................................................. 55
4.5.5 Analog to digital conversion ...................................................................................................... 56
4.4.6 Performance analysis ................................................................................................................. 57
Chapter 5..................................................................................................................................................... 62
Conclusion and future work ........................................................................................................................ 62
5.1 Conclusion ......................................................................................................................................... 62
5.2 Limitations......................................................................................................................................... 63

IV
 5.3 Future work ....................................................................................................................................... 63
References .................................................................................................................................................. 64
Appendix ..................................................................................................................................................... 68
Circuit schematics ....................................................................................................................................... 68

V
List of Figures
Figure 1.1: Processed display of an ECG signal [5] .............................................................................. 3
Figure 1.2: Power spectral plot of EMG signal [7] ............................................................................... 4
Figure 1.3: 10-20 standard for electrode placement in EEG measurement [11] ............................. 5
Figure 1.4: Standard electrode placement for EMG measurement [10] ........................................ 6
Figure 1.5: Generalized architecture of a bio-potential signal acquisition system design [22] ....... 7
Figure 3.1: Flow chart of methodology ........................................................................................ 18
Figure 4.1: Minified internal schematic of INA819 [23] ............................................................... 27
Figure 4.2: High level block diagram .......................................................................................... 33
Figure 4.3: Time domain analysis of EEG source ....................................................................... 35
Figure 4.4: Frequency domain analysis of EEG source ................................................................. 35
Figure 4.5: Time domain analysis of EMG source ....................................................................... 35
Figure 4.6: Frequency domain analysis of EMG source ................................................................ 36
Figure 4.7: Circuit configuration of INA819 ................................................................................ 37
Figure 4.8: Block level analysis of ADS1299 ................................................................................ 39
Figure 4.9: Overvoltage protection circuit ..................................................................................... 41
Figure 4.10: INA819 circuit schematic .......................................................................................... 42
Figure 4.11: Wilson central circuit for referencing voltages of unipolar ECG ............................ 42
Figure 4.12: Daisy chain configuration for ADS1299 ................................................................... 44
Figure 4.13: 3D Layout of the PCB design .................................................................................... 47
Figure 4.14: Bio-potential signal input ........................................................................................ 47
Figure 4.15: Bio-potential signal output ............................................................................................ 48
Figure 4.16: Time domain analysis of power line interference noise ............................................ 49
Figure 4.17: Frequency domain analysis of power line interference noise ................................. 49
Figure 4.18: Time domain analysis of generated white noise ...................................................... 49
Figure 4.19: Frequency domain analysis of generated white noise ............................................. 50
Figure 4.20: Time domain analysis of signal input with added noise signals ............................. 50
Figure 4.21: Frequency domain analysis of input with added noise signals ................................ 50
Figure 4.22: Time domain analysis of amplified output ................................................................ 51
Figure 4.23: Frequency domain analysis of amplified output ........................................................ 51
Figure 4.24: CMRR plot of distinct gain levels ............................................................................. 52

VI
Figure 4.25: PSRR plot for distinct gain levels ............................................................................ 52
Figure 4.26: Snap shot of serial data captured through a serial monitor ........................................ 53
Figure 4.27: Reconstructed serial stream ....................................................................................... 53

VII
List of Tables
Table 1.1: Frequency characteristics of EEG signals ....................................................................... 2
Table 1.2: Average amplitudes of independent ECG waves ......................................................... 3
Table 1.3: Standard electrode placement for a 12-lead ECG measurement .................................. 6
Table 4.1: Design consideration in regards to IFCN standards ................................................... 24
Table 4.2: Component survey for signal amplification ............................................................... 28
Table 4.3: Component survey of analog to digital converter ....................................................... 31
Table 4.4: Summary specification of EEG sample ...................................................................... 34
Table 4.5: Summary specification of EMG sample ..................................................................... 34
Table 4.6: Simulation profile of INA819 ...................................................................................... 38
Table 4.7: Simulation specification for ADS1299 ....................................................................... 39
Table 4.8: Power and grounding requirement of core components ............................................. 45
Table 4.9: Summary of layers with functions .............................................................................. 45
Table 4.10: Minimum trace spacing ............................................................................................ 46
Table 4.11: Feature summary of cyton board .............................................................................. 54
Table 4.12: Summary of ADS1299 based publications ............................................................... 54
Table 4.13: Ching-sung wangs’ 32 Channel EEG acquisition system design feature summary . 55
Table 4.14: Comparison of proposed design with IFCN standards .............................................. 57

VIII
List of Acronyms
ADC Analog to digital converter
A/D Analog / Digital
BCI Brain Computer Interface – used to describe applications for the detection, recording
and analysis of brain wave
CAD Computer Aided Design
CAE Computer Aided Engineering
CAM Computer Aided Manufacturing
CLK Clock – used for interface and data communication clock signals
CMRR Common Mode Rejection Ratio
CS Chip Select – signal for initiating device for data communication
DIN Data in – port used for receiving digital signal
DOUT Data out – port used for transmitting digital signal
DRDY Data Ready – output signal that transition from high to low indicating new conversion
data are ready
DSP Digital Signal Processor
ECG Electrocardiogram – diagnostic tool for assessment of electrical and muscular function
of the heart
EEG Electroencephalography – a diagnostic procedure to detect abnormalities in brain
waves
EMG Electromyogram – diagnostic procedure to detect muscular abnormalities
edf European Data Format – used for storing time series data
GPO General Purpose Output – pin used for general purpose of digital input and/or output
GUI Graphic User Interface
HPF High Pass Filter
IC Integrated Circuit
IFCN International Federation of Clinical Neurophysiology – reference for set of standards
in brain-computer applications
INA Instrumentation amplifier
LNA Low Noise Amplifier
LPF Low Pass Filter

IX
MISO Master In Slave Out
MOSI Master Out Slave In
MUX Multiplexer
OSR Over Sampling Ratio – design parameter for sigma delta converters
PCB Printed Circuit Board
SCLK Serial Clock
sEMG Surface Electromyogram
SNR Signal to noise ratio
SOC System on chip
SPI Serial Peripheral Interface - a full-duplex synchronous serial interface for connecting
low-/medium-bandwidth external devices using four wires
SPICE Simulation Program with Integrated Circuit Emphasis

X
 Chapter 1
Introduction
1.1 Overview
Bio-potentials signals originate from the human body as occurrence of potential differences
between compartments [1]. These signals are generated due to the electrochemical activity of
certain class of cells that are components of the nervous, muscular or glandular tissue. The
electrical activity of each cell is described by the ion exchange through the cell membrane. These
signals possess various properties and characteristics that contribute to their diagnostic value. The
available bio-signals are grouped into two major categories which are: bio-electric signals (i.e.
ECG, EEG and EMG) and bio-acoustic signals (i.e. lung sounds, hears sounds). These signals are
tiny carriers of information that provide crucial data about the bodily functions and are accurate
indicators of the physiological and neurological state of human beings.

In a nutshell, an EEG, EMG and ECG are the physical measurements of the signals generated by
the human body, human muscle and human heart respectively [2]. These organs are filled with
mobile ions such as calcium and potassium which constantly flow through the body. Through
accurate and highly sensitive detection mechanism, these signals can be retrieved and interpreted.
These signals, if interpreted correctly, can give extensive explanation to the physical functioning
state of the body, which is very crucial in the diagnosis of illnesses related to the organs.

The weakness of these signals in amplitude and frequency have put enormous challenges in
designing systems that accurately measure them. Some of the most predominant challenges faced
in designing such systems are: Power line interference, Instrumental Imprecisions, DC bias
voltage, Gain adjustability, Digital System consideration and circuit complexity [2].

In the last few decades, consumer grade bio-medical electronics have gone through numerous
changes. Initially, these devices were to be operated only by licensed physicians under certain
protocols. However, the advancement in research within this area has paved the way for the public
to put their hands on such devices. These devices usually take the form of a portable hardware
such as a headset or an armband, and usually have the means of not just processing the data but to
send it over some form of communication channels. These bio-medical electronics provide a
diverse range of functionality from detection of brain signals and heart rate, to digital storage and

1
interactions through mobile applications as well as web interfaces. However, these electronic
devices cannot be utilized for medical diagnosis purposes as they do not meet international bio-
medical standards.

The aim of this thesis is to provide a solution to the critical problems pertaining to noise
cancellation, digital filtering, circuit complexity, availability and affordability by designing a data
acquisition hardware that can be used for the acquisition of EEG, ECG and EMG signals that meets
international medical standards. The project proposes a step by step approach of bio-signal
processing from detection through to data communication under clearly defined stages of signal
conditioning, analog to digital conversion and data communication.

1.2 Theoretical background and design principles

1.2.1 Signal characteristics
EEG, ECG and EMG signals have broad classification independently with very similar
characteristics to each other which is discussed in the coming sections. However, the property that
is most common to these signals is the fact that they are very weak in amplitude and low in
frequency. This property makes these signals highly susceptible to noise interference which have
numerous sources but can be generally categorized as [3]: personal-related and technical. Personal-
related noise sources refer to artefacts that arise from bodily movement of individual under study,
sweating, breathing and so on. Technical artefacts refer to artefacts such as AC power line noise,
fluorescent lighting, electrical machinery and computers.

A. EEG signals

Different mind states lead to different EEG displays. The four main mind states—alertness, rest,
sleep, and dreaming—have associated brain waves named beta, alpha, theta, and delta
respectively [4]. Aside from providing crucial information in regards to the state of the mind, these
signals are key indicators of any illness or abnormality in regards to the brain. The brain wave
patterns are classified according to their frequency which is described in table 1.1.

2
 Table 1.1: Frequency characteristics of EEG signals
No. Signal Name Frequency (Hz)
1 Delta <4
2 Theta 4-7
3 Alpha 8 - 15
4 Beta 16 - 31
5 Gamma > 32
6 Mu 8 - 12

B. ECG signals

Relatively speaking, ECG signals are much stronger than EEG signals and hence require less
amplification. The useful frequency band of ECG signals range from 0.5 – 100Hz. One cardiac
cycle of ECG signal consists of the P wave, QRS complex along with T waves [6]. P wave
represents depolarization and the QRS represents ventricular depolarization. T wave represents
rapid repolarization of the ventricles. The signal representation under normal scenario is shown
in the figure 1.1.

Figure 1.1: Processed display of an ECG signal [5]

The dynamic range in amplitude of the segments vary from each other as given in table 1.2 [5]

3
 Table 1.2: Average amplitudes of independent ECG waves
No. Wave Amplitude
1 P wave 0.25 mV
2 R wave 1.6 mV
3 Q wave 25% of R wave
4 T wave 0.1 – 0.5 mV

C. EMG signals

EMG signals refers to the recording of the potential difference resulting from muscle movement
[3]. Relatively speaking, EMG signals possess stronger amplitude and wider bandwidth compared
to EEG and ECG signals. The amplitude of EMG signal is stochastic (random) and can be
reasonable represented by a Gaussian distribution function. The amplitude can range from 0 to 10
mv (peak-to-peak) or 0 to 1.5 mv (rms).

The power density spectrum of the EMG signal ranges from zero to four hundred Hertz for
many muscles [3]. Above this frequency, the amplitude of the signal is less than 1µV rms and are
no longer distinguishable from the noise of the detection and recording system. There are some
exceptions, such as the masseter muscle, where the frequency distribution reaches up to 600 Hz.
As can be inferred from figure 1.2, the dominant energy of EMG signal lies in the range of 0 – 200
Hz. More specifically, the frequency range of 0 – 150 is one that is the most usable for preliminary
diagnosis purposes.

Figure 1.2: Power spectral plot of EMG signal [7]

4
1.2.2 Signal measurement type and standards
1.2.2.1 Measurement type
When considering measurement types in clinical EEG, EMG and ECG measurement, there are
two general categories that are employed: invasive and non-invasive [3]. Invasive EEG, ECG and
EMG recordings refer to measurements that require surgery. This type of measurement requires
highly trained physicians under specially equipped laboratories and centers. It requires the surgical
penetration of the scalp, heart and muscle of the understudy and has to be done with the utmost
care and expertise. On the other hand, there is the non-invasive method of bio-potential signal
measurement which requires no surgical procedure and is done through the use of surface
electrodes. The measurement type proposed in this work is a non-invasive type carried out with
pre-gelled gold-plated electrodes.

1.2.2.2 Measurement standards

EEG, ECG and EMG signals have different types of measurement standards. When selecting
which type of standard to employ to use for various types of bio-potential signals, the critical point
to consider is the expected detail of the recording required. For the measurement of brain waves,
the widely accepted standard electrode placement is known as the 10-20 system which is a method
to describe and apply the location of scalp electrodes in the context of an EEG exam,
polysomnography sleep study, or voluntary lab research [6]. This type of electrode placement is
depicted in figure 1.3.

Figure 1.3 10-20 standard for electrode placement in EEG measurement [11]

5
For the measurement of EMG signals, the placement of electrodes depends on the type of
application required. Generally, there will be more than one measurement electrode placed on the
muscle area under inspection [20]-[21]. One of the electrodes will be used as a reference so that
the measurement starts from zero. Other electrodes will be used for the measurement of the muscle
potential between the muscle ends of placement. Sample placements is shown figure 1.4.

Figure 1.4 Standard electrode placement for EMG measurement [10]

In regards to ECG measurement, the scheme that yields detailed information is known as the 12-
lead ECG [5]. It is composed of twelve signals or ‘leads’ measured from the limbs and six
positions on the chest called precordial. The precordial (1/2 of the signals) are measured as the
potential difference between each exploring electrode located on the chest, and an assumed
constructed ‘zero’ reference. This ‘zero’ reference was introduced by F. N. Wilson in 1931 and
named after him as Wilson’s Central Terminal (WCT). The placement of the electrodes is given
in the table 1.3.

6
 Table 1.3 Standard electrode placement for a 12-lead ECG measurement
Electrode name Electrode Placement
RA Right arm
LA Left arm
RL Right leg
LL Left leg
V1 Right of the sternum (between ribs 4 and 5)
V2 Left of the sternum (between ribs 4 and 5)
V3 Between leads V2 and V4
V4 Mid-clavicular line
V5 Left anterior axillary line
V6 Mid-axillary line

1.2.3 Data acquisition system architecture

When designing bio-potential data acquisition systems, engineers take diversified approaches in
regards to sensing mechanisms, noise cancellation, filter designs as well as digital system
development. The approaches taken in designing such systems is made in regards to what type of
optimization is required. However, the general design process of bio-potential signal acquisition
system can be generalized through a high-level system architecture given in figure 1.5.

Signal
Conditioning

Signal PC
Conditioning
ADC Memory
Signal MUX
Conditioning Analysis
TX Software

TX RX TX
Sensor
Signal Display
Wireless Conditioning
TX
Figure 1.5 Generalized architecture of a bio-potential signal acquisition system design [22]
TX
7
The main components of a DAQ are shown in figure 3.5. The inputs are from sensors that are
usually connected by cables to signal conditioning circuits to prepare their outputs to be digitized.
Signal conditioning includes amplification, filtering for noise, level shifting, or other corrections.
In some applications the sensor may be far from the DAQ system in which case a wireless link can
be established using Wi-Fi or other wireless technology [22].

The processed sensor signals then go to a multiplexer (MUX) that selects the sensor to be digitized
and passes the signal along to an analog-to-digital converter (ADC) that samples the analog signal
from the sensor and converts it to a stream of binary values. Sampling rates are usually low, from
several samples per hour to 1 MS/s. The resulting data words are then sent to a PC and stored in a
file. The data can then be analyzed by software, used in analysis or converted into appropriate
displays or graphs [22].

1.2.4 Safety and standard

Medical devices are subject to strict general controls and procedural regulations [3]. The
development and use of standards are vital to ensuring the safety and efficacy of medical
devices. Numerous regulatory agencies and standards organizations collaborate to establish the
accepted standards for medical equipment. Standard‐setting activities include the development of
performance characteristics, characterization and testing methodologies, manufacturing practices,
product standards, scientific protocols, compliance criteria, ingredient specifications, labeling, or
other technical or policy criteria. However, it should be noted that the type of standard to be
followed is defined by the target market of the equipment as the standards are required to be met
based on a country basis.

Aside from general standards, patient safety is also a critical component of the design process. As
EEG, ECG and EMG components are directly attached to the subject under test, the system design
should be done with special emphasis on the power circuitry and protection options.

Moreover, for the case of EEG data acquisition design, the International Federation of Clinical
Neurophysiology (IFCN) published a set of recommended practices for EEG recording [9]. Here
a Workgroup of IFCN experts presents unanimous recommendations on the following procedures

8
relevant for the topographic and frequency analysis of resting state EEGs in clinical research
defined as neurophysiological experimental studies carried out in neurological and psychiatric
patients:

• Recording of resting state EEGs (environmental conditions and instructions to participants;

montage of the EEG electrodes; recording settings);
• Digital storage of resting state EEG and control data;
• Computerized visualization of resting state EEGs and control data (identification of
artifacts and neuropathological resting state EEG waveforms);
• Extraction of "synchronization" features based on frequency analysis (band-pass filtering
and computation of resting state EEG amplitude/power density spectrum);
• Extraction of "connectivity" features based on frequency analysis (linear and nonlinear
measures);
• Extraction of "topographic" features (topographic mapping; cortical source mapping;
estimation of scalp current density and dura surface potential; cortical connectivity
mapping)
• Statistical analysis and neurophysiological interpretation of those resting state EEG
features

1.2.4 Summary

EEG, ECG and EMG signals are bio-potentials that possess very weak amplitude and narrow
frequency. The proper acquisition, recording and analysis of these signals yields crucial diagnostic
information of subjects and can be the basis of further studies in regards to bodily functions. There
are different types of methods that can be employed for the measurement of these signals.
However, some of the widely applied measurement standards are: 10-20 system for EEG
measurement; 12-lead for ECG recording; differential placement for EMG. The measurement
standards employed are dependent on the required accuracy and diagnostic detail.

Consequently, in the design of data acquisition systems for bio-potential signals, modularizing the
entire approach for better understanding of the architecture is necessary. Accordingly, the major
sections of the systems can be broken down as sensing, conditioning, analog-to-digital conversion

9
and digital signal processing. Hence, the design of EEG, ECG and EMG data acquisition systems
involve critical consideration: sensing methodology; signal amplification; noise removal; analog-
to-digital conversion architecture as well as mechanisms involved; digital signal processing
requirements and specifications. Moreover, study of general patient safety requirements,
regulations as well as useful recommendations as well as standards should be thoroughly
understood and utilized.

1.3 Problem statement

In the current medical provision of low-income countries, getting a brain scan, a thorough heart
examination, a physiotherapy, or related diagnoses have become a luxury for reasons such as:

• Cost of examination
• Shortage of equipped medical centers
• Large queues
• Length of appointment
• Shortage of qualified physicians equipped with the diagnostic knowledge of scans
• Absence of cost-effective supplementary solutions.

Bio-medical equipment manufacturers across the globe come out with new products that minimize
human error whilst providing impeccable performance and patient safety. However, the costs of
these equipment are very high and their functionality is mainly limited to a single bio-potential
signal application. Aside from the product spectrum of bio-medical equipment, research institutes
are also actively involved in the improvement of circuit design of such systems in regards to noise,
circuit complexity and digital filtering. Moreover, there are numerous ongoing BCI researches that
are aimed at advancing the field in regards to flexibility and accuracy.

Evidently, Researchers tend to provide special emphasis on the development and advancement of
BCI which mainly revolves around the human brain while products are restricted to single bio-
potential signal applications. However, when carefully studying the signal characteristics of the
brain, heart and muscles of the human body, it becomes evident that these signals possess highly
similar traits in regards to signal amplitude as well as frequency. Hence, this thesis provides a
supplementary solution by designing a hardware capable of data acquisition of EEG, ECG and
EMG signals with optimization in regards to complexity and performance.
10
1.4 Objectives
1.4.1 General objective
The main objective of this research is to design a hardware for the measurement of EEG, ECG and
EMG bio-potential signals.

1.4.2 Specific objectives

• Design of a bio-medical electronic circuit with optimized circuit complexity, performance
and modularity for application in a 3-in-1 EEG, ECG and EMG bio-potential data
acquisition
• Carry out a detailed market research in integrated circuits manufactured for medical
instrumentation
• Develop a simulation environment that mimics real world application scenario and carry
out verification via simulation
• Analyze, interpret and provide valid conclusions and recommendation for future studies as
well as adaptation into a feasible product

1.5 Scope of the study

The scope of this thesis is focused on the design and simulation of a data acquisition hardware for
EEG, ECG and EMG signal application. Although highly accurate simulation models and tools
are used, due to limitation in resource and time, experimental verification of the design isn’t carried
out. Moreover, the thesis is strictly concerned with the hardware part of the data acquisition and
doesn’t account for the software counterpart of the design.

1.6 Significance of the study

The ultimate outcome of this study has impacts in regards to both research and product
development. The final design of the PCB is done so as to allow actual manufacturing which
tackles the major problem of medical provision in low-income countries by acting as a
supplementary solution. Furthermore, the digital counterpart can be developed simultaneously and
scaled up to a nation-wide repository for medical data storage, remote monitoring and diagnosis.

Moreover, the study also contributes to previous researches that have worked on optimization in
regards to complexity and overall performance. It also extends the concept of a single bio-potential

11
data acquisition hardware being used for multiple purposes, which has been suggested by previous
studies, by designing a bio-potential data acquisition hardware that can be used for three distinct
purposes i.e. EEG, ECG and EMG.

1.7 Thesis outline

Included in this research is the design, simulation and analysis of a hardware system for EEG,
ECG and EMG signal acquisition. Accordingly, the thesis is organized with five distinct chapters
outlined as follows.

Chapter one provides a brief outline in regards to bio-potential signals, their significance as well
as current stages of bio-medical equipment design. Problems residing in the area of study is
discusses along with the objectives of the research. Finally, the scope of the study along with the
significance in regards to research as well as product is provided.

Chapter two provides a literature review related to the research at hand. The review is done so as
to provide a seamless continuation of concepts by initially providing reviews in regards to signal
characteristics of the bio-potential signals followed up by previous researches in the area of bio-
potential data acquisition hardware design. Finally, review in regards to common design
methodologies and patterns is carried out.

Chapter three deals with the methodology employed in the research by describing the methods
employed, tools utilized and research process flowchart. This chapter is meant to provide a detailed
analysis of the steps followed in data collection, design, simulation and analysis used in the
research.

Chapter four goes through the entire circuit design in detail through the application of the concepts
discussed in chapter three through an in-depth circuit design and simulation. Once the design is
verified through simulation, the PCB design is also carried out after which, results and discussions
are carried out.

Finally, chapter five provides a conclusion based on analysis of discussion carried out in chapter
four. Moreover, it contains what the future holds for this research as well as shortcomings that
need improvement in its future endeavors.

12
 Chapter 2
Literature review
2.1 Bio-potential signals and measurement practices
Wim V. Drongelen [1] describes biopotentials as signals of the body that originate within
biological tissue as potential differences occurring between compartments. Generally, the
compartments are separated by a (bio)membrane that maintains concentration gradients of certain
ions via an active mechanism (e.g., the Na+/K+ pump). Moreover, John W. Clark Jr [2] sets out
in detail the origin of bio-potentials commencing from a discussion at the cellular level. In his
work, he identifies a class of cells known as excitable cells that are components of nervous,
muscular or glandular tissue. Electrically, they exhibit resting potential and when properly
stimulated, an action potential.

Measurement of biopotential signals through the utilization of specialized electrodes is discussed

in [3]. The function of recording electrodes is described as the coupling of the ionic potentials
generated inside the body to an electronic instrument. Furthermore, the classification of
biopotential electrodes is done as noninvasive (skin surface) or invasive (e.g., microelectrodes or
wire electrodes). Invasive bio-potential signal recordings refer to measurements that require
surgery. This type of measurement requires highly trained physicians under specially equipped
laboratories and centers. It requires the surgical penetration of the scalp, heart and muscle of the
understudy and has to be done with the utmost care and expertise. On the other hand, there is the
non-invasive method of bio-potential signal measurement which requires no surgical procedure
and is done through the use of surface electrodes.

Biopotential measurements must be carried out using high-quality electrodes to minimize motion
artifacts and ensure that the measured signal is accurate, stable, and undistorted. Body fluids are
very corrosive to metals, so not all metals are acceptable for biopotential sensing. Furthermore,
some materials are toxic to living tissues. For implantable applications, we typically use relatively
strong metal electrodes made, for example, from stainless steel or noble materials such as gold, or
from various alloys such as platinum-tungsten, platinum-iridium, titanium-nitride, or iridium-
oxide. These electrodes do not react chemically with tissue electrolytes and therefore minimize
tissue toxicity. Unfortunately, they give rise to large interface impedances and unstable potentials.

13
External monitoring electrodes can use nonnoble materials such as silver with lesser concerns
of biocompatibility, but they must address the large skin interface impedance and the unstable
biopotential. Other considerations in the design and selection of biopotential electrodes are
cost, shelf life, and mechanical characteristics [3].

2.2 Electroencephalogram
Electroencephalogram (EEG) is the recording of electrical activity along the scalp [3]. It further
elaborates the recording as the measures of voltage fluctuations resulting from ionic current flows
within the neurons of the brain. In clinical contexts, EEG refers to the recording of the brain’s
spontaneous electrical activity over a short period of time, usually 20-40 minutes, as recorded from
multiple electrodes placed on the scalp. Diagnostic applications generally focus on the spectral
content of EEG which refers to the type of neural oscillations that can be observed in EEG signals.

EEG is most often used to diagnose epilepsy, which causes obvious abnormalities in EEG
readings. It is also used to diagnose sleep disorders, coma, encephalopathy, and brain death. EEG
is used as a first-line method of diagnosis for tumors, stroke and other focal brain disorders, but
this use has decreased with the advent of high-resolution anatomical imaging techniques such as
MRI and CT. despite limited spatial resolution, EEG continues to be a valuable tool for research
and diagnosis, especially when millisecond-range temporal resolution (not possible with CT or
MRI) is required [3].

The system of EEG measurement involves hooking up several pairs of electrodes on a patients’
head. These electrodes are disks that conduct electrical activity, capture it from the brain and
convey it out through a wire to a machine that amplifies the signal. Electrodes attached in pairs
on the head, measure the difference in voltage between the pairs. The measured electrical activity
of brain waves is correlated to a persons’ state of mind and the brain patterns form
wave shapes that resemble sinusoids. Usually, they are measured from peak to peak and
normally range from 0.5 to 100µV in amplitude [4].

2.3 Electrocardiogram
The registration of the ECG signal Known as electrocardiogram, represents the recording of the
electrical potential of the heart. Physicians record ECG signal easily and noninvasively by
attaching small electrodes to the human body. Electrocardiogram is the standard tool to diagnose
14
the heart disease. While diagnosis the different artifacts get introduced in the ECG signal like
Electrode contact noise, motion artifacts, base line drift electrosurgical noise, and power line
interference [4].

Relatively speaking, ECG signals are much stronger than EEG signals and hence require less
amplification. The useful frequency band of ECG signals range from 0.5 – 100Hz. One cardiac
cycle of ECG signal consists of the P wave, QRS complex along with T waves [5]. P wave
represents depolarization and the QRS represents ventricular depolarization. T wave represents
rapid repolarization of the ventricles.

Pertaining to signal measurement, the ECG measurement scheme that yields detailed information
is known as the 12-lead ECG [5]. It is composed of twelve signals or ‘leads’ measured from the
limbs and six positions on the chest called precordial. The precordial (1/2 of the signals) are
measured as the potential difference between each exploring electrode located on the chest, and an
assumed constructed ‘zero’ reference. This ‘zero’ reference was introduced by F. N. Wilson in
1931 and named after him as Wilson’s Central Terminal (WCT).

2.4 Electromyogram
An electromyogram (EMG), is a graphical recording of electrical activity within muscles.
Activation of muscles by nerves results in changes in ion flow across cell membranes, which
generates electrical activity. This can be measured using surface electrodes placed on the skin
over the muscle of interest [1].

The EMG signal is the electrical manifestation of the neuromuscular activation associated with a
contracting muscle. It is an exceedingly complicated signal which is affected by the anatomical
and physiological properties of muscles, the control scheme of the peripheral nervous system, as
well as the characteristics of the instrumentation that is used to detect and observe it. Most of
the relationships between the EMG signal and the properties of a contracting muscle which are
presently employed have evolved serendipitously. The lack of a proper description of the EMG
signal is probably the greatest single factor which has hampered the development of
electromyography into a precise discipline [39].

15
Measurement of surface EMG is dependent on a number of factors and amplitude as the surface
EMG signal varies from the microvolts to the low millivolts range. The amplitude and time and
frequency domain properties of the EMG signal are dependent on factors such as [3]: Timing and
intensity of muscle contraction; Distance of the electrode from the active muscle area; Properties
of the overlying tissue; Electrode and amplifier properties; Quality of electrode – skin contact.

If properly integrated, EMGs can be used to detect abnormal electrical activity of muscle that can
occur in many diseases and conditions, including muscular dystrophy, inflammation of muscles,
pinched nerves, peripheral nerve damage (damage to nerves in the arms and legs), amyotrophic
lateral sclerosis (ALS), myasthenia gravis, disc herniation, and others [3].

2.5 Related works

There are numerous works pertaining to EEG, ECG and EMG applications. Even though the
majority of these researches are restricted to single signal applications and tend to push boundaries
in regards to technological advancements, the considerations taken in regards to the design is
extremely helpful for any designs that share the scope of application

In regards to equipment design pertaining to EEG, ECG and EMG signal acquisition. the primary
focus of biomedical signal processing was on filtering signals to remove noise [4]. These sources
of noise include the interference from power lines, instrumental imprecisions and more. Hence,
the major objective during these days was to implement powerful noise cancellation mechanisms
specifically designed for medical applications. The widely used method is through the use of
frequency suppression while the other method is through statistical signal processing.

There are abundant researches done in the area of bio potential signal acquisition system that
approach the idea from a diversified point of view. Some of the works include: multichannel data
acquisition for brain control interface [14], Brain-computer interfaces for communication and
control [40], past, present and future of BCI [41] and more. However, most researches are
restricted to detection and analysis of a single type of bio potential signal acquisition and restricted
to non-clinical applications.

A detailed multichannel signal acquisition system is designed in [14] where a successful design of
a 32 channel EEG system is developed for the purpose of brain control interface application. The

16
research takes into account existing standardization in regards to brain wave applications and
provides a portable hardware with a design that improves single-power AC-coupled circuit, which
effectively reduces the DC bias and improves the error caused by the effects of part errors. The
work also provides the software counterpart for the display.

Design considerations for mixed signal application systems which includes grounding, choice of
power supply, power supply filters, low pass filters for ADC to remove wideband noise and after
DACs for the reconstruction of the required analog signals is put forth in [34]. These considerations
have considerable utility when physical prototypes are built.

In [16] and [17], signal acquisition system for the application in the area of surface electromyogram
is carried out. These works present a detailed topology for the detection of muscle signals based
on an SOC specifically designed for bio-potential signal applications. The works present a four-
to-eight channel data acquisition system with variable sample rate for the application in surface
electromyogram and electroencephalogram.

Another emerging, important need in a number of biomedical experiments is classification. In

clinics, the goal of classification is to distinguish pathology from normal [41]. For instance,
monitoring physiological recordings, clinicians judge if patients suffer from illness; watching
cardiac MRI scans, cardiologists identify which region the myocardium experiences failure;
analyzing gene sequences of a family, geneticists infer the likelihood that the children inherit
disease from their parents. The rigorous research field of biomedical engineering is in great
demand of the integration of bio-medical devices with statistical generation and artificial
intelligence.

In [31], one of the first multi-person non-invasive direct brain-to-brain interface for collaborative
problem solving is designed. The interface combines EEG to record brain signals and transcranial
magnetic stimulation to deliver information noninvasively to the brain. The design is carried out
through the use of a commercial EEG machine along with Arduino boards to carry out customized
tasks. The interface allows three human subjects to collaborate and solve a task using direct brain-
to-brain communication. The design also incorporates careful consideration of typical server-client
TCP protocol to transmit information between computers for the purposes of record, analysis and
display.
17
An experiment utilizing OPENBCI board for the collection of data from hand gestures for the
purpose of gesture identification is done in [32]. EMG signal is collected through the use of three
individual electrodes placed on the forearm of the subject and transferred through a single channel.
Butterworth bandpass filter is used to extract the signal of choice after which an algorithm based
on the Hilbert transform is used for identifying the dynamic threshold and find the action segment.

In regards to the benefit of BCI technological advancements in clinical applications, the prospect
of improving the lives of countless disabled individuals through a combination of BCI technologies
is carried out in [35]. In doing so, four application areas where disabled individuals could greatly
benefit from advancement in the area are identified as: communication and control, motor
substitution, entertainment and motor recovery. Moreover, existing state of the art equipment and
future development are assessed whilst discussing the main research issues in the identified
spectrum.

Furthermore, in [33] – [37], diversified researches based on ADS1299 are carried out. These
researches span the areas of EEG, ECG and EMG application for numerous purposes that take
advantage of the fact that the integrated circuit has been specifically designed for clinical
application that require specific performances in regards to low signal applications.

2.7 Summary
EEG, ECG and EMG signals are bio-potential signals whose measurement and analysis yields
critical information in regards to bodily function. Measurement of these signals can be done
through an invasive method which requires surgery or through a non-invasive method through the
use of highly sensitive surface electrodes is widely used. The type of electrode placement depends
on the diagnostic required measurement detail as well as expected analysis.

Moreover, through the analysis of the literatures discussed, it becomes evident that most of the
researches are concerned with the design of signal acquisition intended for single bio-potential
signal application. Moreover, researches in the area of EEG application are focused on the
extension of BCI applications and lack the clinical application for basic medical provision.
Furthermore, the literatures focused on the signal analysis and description are proof that EEG,
ECG and EMG signals possess highly similar qualities in regards to their amplitude as well as
frequency. Hence, designing a 3-in-1 system for multipurpose application is not farfetched.
18
 Chapter 3
Methodology
3.1 Design and implementation procedure
The approach devised for this research is shown in figure 3.1. It involves seven stages which are
subsequently brought together to meet the objectives of the research.

Start

Problem formulation

Identification of objectives

Definition of thematic scope

No
Preliminary

Problems
addressed

Yes

Figure 3.1 Flow chart of methodology

19
 A

Analysis of bio-potential signals

Review and analysis of data

acquisition system design,
performance evaluation criteria &
standards

Review and analysis of related works

Literature review & analysis

System architecture development

Component research, identification

and implementation

System design

Figure 3.1 (continued)

20
 B

Identification of low-potential signal

sources

Development of PSPICE-Simulink co-

simulation

Simulation and data recording

Simulation

No Design
Expected result
enhancement

Simulation and data recording

Simulation and data recording

Hardware design

Figure 3.1 (continued)

21
 B

Result of recorded data

Documentation of end-to-end design

and simulation

Analysis and conclusion

Analysis and evaluation

Start

Figure 3.1 (continued)

3.1.1 Preliminary
The preliminary stage of the methodology is concerned with the identification of residing problems
of the thematic area in regards to both research as well as product. It lays the foundation of the
research by clearly defining the objectives which are meant to solve residing issues in the area.
Moreover, the scope of the research is discussed which defines the overall expectation of the end
output.

3.1.2 Literature review and analysis

This phase is concerned with the review of publications related to the research at hand for the
purposes of better understanding and conception of the research topic as well as determine further
information. The extraction is made through the use of books, articles, journals, conference papers,
online portals and more. Furthermore, the use of literature review and analysis is done through the

22
definition of two distinct stages: Study and analysis of bio-potential signals; Common practices in
data acquisition system design.

During the first phase of literature review and analysis, an in-depth study and analysis of bio-
potential signals is carried out. The study addresses the signal characteristics of EEG, EMG and
ECG signals in regards to their amplitude, frequency, diagnostic behavior, measurement measures
etc. Furthermore, comparative analysis of the signals in regards to their similarities and differences
are carried out.

The second phase is concerned with the study of data acquisition system design which includes
the study and analysis of diversified approaches of previous researches in the thematic area.
Moreover, the identification of critical performance criteria mentioned in previous researches as
well as standards and practices are taken into consideration for the purpose of drafting optimized
metrics for this research.

3.1.3 System design

The system design phase of the research is done through two distinct stages which are: system
architecture development; component identification and implementation. In the system
architecture development, the identification of system blocks which is concerned with undertaking
a block level analysis of the system from signal sensing through to digital output. Moreover,
performance requirement development is done in order to set the goal and metrics of the end design
through the aid of analysis carried out on previous researches.

Once the system architecture development is complete, the second stage deals with the
identification of circuit components based on a pre-defined selection metrics. The circuit
components are selected based on the criteria through an intensive market search from IC
manufacturers extensive database. Moreover, a component survey record is also done for the
purpose of future reference.

3.1.4 Simulation
This phase is concerned with the development of a simulation environment for the purposes of
design evaluation, correction, validation and analysis. The simulation is done with a PSpice –
SIMULINK co-simulation through which the data acquisition system designed is tested end-to-

23
end. Furthermore, the output of the simulation is thoroughly analyzed through the reconstruction
of the digital output stream. The simulation is inclusive of the following:

• Actual bio-potential records from online database as signal sources

• IC models from manufactures for individual design blocks
• Performance evaluation of critical components through the additions of noise

3.1.5 Hardware design

In the hardware design phase, the PCB development of the validated design is carried out. The
design is carried out with the aid of CAE tools and includes the following:

• Circuit schematic
• PCB design
• Routing and simulation

3.1.6 Analysis and evaluation

In this phase, an in-depth assessment of simulation results is carried out. Moreover, a comparative
analysis of the overall design against previous researches as well as existing product is done. The
analysis is done against the following key points:

• Existing open source product

• Application areas
• Hardware property
• General performance

3.1.7 Conclusion
The last phase is concerned with providing a conclusion based on discussions in previous phases,
obtained results and overall output. Moreover, suggestion in regards to the future of the research
along with its limitation is carried out.

3.2 Tools and signal resources

For the purpose of overall circuit design, simulation and analysis, carefully selected Computer
aided engineering (CAE) tools are utilized. For the entirety of the project, the tools employed with
their scopes are outlined below:

24
• Cadence OrCAD V16.3 CAE software package is used to design the electrical schematic,
PCB layout and routing
• PSpice AD V16.3 is used for the simulation of the LNA stage of the design. Moreover, it
is used detailed analysis and comparison in regards to variable amplification, CMRR and
PSRR
• MATLAB R2018a is used to generate a script for the purpose of creating a user setting for
data supply to GUI based simulation done on Simulink
• Simulink is used for the simulation and analysis of the analog-to-digital conversion stage.
Furthermore, the simulation of data transmission through SPI and the reconstruction is
done.
• DSToolbox is used parameter settings such as: OSR, NTF synthesis and coefficient
mapping
• SIMSIDES is used to add non-idealities to ADC simulation block such as: clock jitter, op-
amp saturation and switch non-linearity
• Physionet.org is used as the source for bio-potential signals. Physionet.org provides
decades of actual records of patients across the globe. The fact that bio-potential signals
resemble low amplitude noises makes it very hard for result analysis. However, the
utilization of actual records contributes to the analysis and authenticity.

25
 Chapter 4
Proposed circuit architecture and implementation
4.1 Overview
There are three distinct signals that are pertinent in this work: EEG, ECG, EMG. Careful analysis
of the individual signals makes the similarity in regards to amplitude and frequency evident.
Hence, the methods employed in regards to the design of the hardware takes into consideration
critical factors such as noise cancellation, amplification, analog-to-digital conversion and digital
signal processing.

The approach gives special attention for the elimination of power line interference, gain
adjustment, anti-aliasing filter requirement, circuit complexity and extended functionality. The
issue of noise cancellation involves the use of high performance LNAs that provide stable common
mode rejection for application specific band of frequency. The selection of an LNA that best fits
the aimed performance is meticulously done through a comparison matrix: Number of channels,
available gain, supply voltage, minimum CMRR, bandwidth and specific application area.

Circuit complexity optimization is done through the integration of a state-of-the-art analog to

digital converter provided by Texas Instruments which eliminates the use of anti-aliasing filters
resulting in the significant reduction of circuit complexity whilst significantly contributing to a
boost in performance.

The hierarchical circuit design is validated through an integrated circuit simulation that takes into
account all non-idealities for employed circuit components such as input referred noise, jitter noise
and component saturation whilst providing a general simulation model for future inferences and
continuations in the sector through a parameter setting environment allowing users to check all the
dynamic variables that are essential for performance evaluation of like systems.

4.2 System design

The circuit topology of the bio-potential signal acquisition system can be categorized into two as
analog sub-system and digital sub-system. The analog sub-system is concerned with the extraction
of the usable band of frequency and amplification of the signals to an operable point. In a circuit
that has an integration of numerous passive elements, the consideration of noise is a very critical
issue. The issue of noise can be further classified into three classes: Power line interference and
26
sources from operation environment; Thermal noise generated by the resistive property of
electrodes; Flicker noise due to components intrinsic behavior. In order to get rid of information
loss due to the presence of these sources of noise, one must come up with an effective method of
noise cancellation. Moreover, the amplification of the signals should be adjustable as the amplitude
vary and component saturation should be avoided.

The digital sub-system on the other hand is concerned with the conversion of the analog signal to
its digital equivalent and transmission of the signals for further processing and display. The key
component in this sub-system is the ADC. When considering the selection of the ADC, the key
points to consider include architecture, resolution, anti-aliasing filter design and signal selection
which have a direct impact on circuit complexity as well as performance.

4.2.1 Goals and Metrics

This research aims at providing an optimized design for bio-potential signal acquisition in regards
to noise cancellation, circuit complexity and standardization. There are several standards that must
be met when designing circuits for bio-medical applications. Specific to this work, there are
standards that must be met respective to the signal under consideration. However, when studying
the properties of the signal, it becomes apparent that the weakest in amplitude and frequency are
EEG signals. Accordingly, this work proposes to use the IFCN standards as the design metrics in
regards to standardization. The IFCN has put forth standards to be met in designing such systems.

Table 4.1 Design consideration in regards to IFCN standards

No. Term IFCN Standard
1 CMRR (dB) ≥110
2 LPF (Hz) ≥70
3 Sample Rate ≥200
4 Bits ≥12
5 Channel ≥24

In section 1.2.2, the measurement standards for ECG, EMG and EEG have been discussed which
is set as the second metric in this work. Previous works in this area have provided meticulously
mapped solutions for individual measurement of bio-potential signals. However, this research
27
takes the measurement type one step forward and provides a detailed measurement practice for all
three types of signal. Accordingly, the hardware tends to provide a detailed 32-channel EEG
reading, 12-lead ECG reading as well as dynamic EMG reading.

4.2.2 Bio-potential signal conditioning

As discussed in section 1.2.3, the high-level architecture of bio-potential signal acquisition is
usually built using a pre-amplifier followed by an LPF and HPF. However, this work proposes the
implementation of an LNA that can be used specifically for bio-potential signal amplification with
high and stable CMRR. This approach tends to eliminate the use of higher order filters of various
topologies to get rid of surrounding noise. Generally, the noise one should be concerned with when
designing such systems are 50/60 Hz power line interference, fluorescent lighting (depending on
the environment of operation), and miscellaneous noises (i.e. computers, machinery etc.).

To find an LNA that can carry out this operation for the intended purpose efficiently, a component
research is carried out. This research has its own measurement grids which are discussed below.

4.2.2.1 Gain adjustment

The circuit is intended to operate for three types of signals that are different in amplitude and
frequency. The critical issue to consider in this case is the concept of IC saturation. If the signals
are amplified by the same gain factor, G, then the output of amplified EEG and EMG for instance
will have different implications. EEG signals are in the range of microvolts which after
amplification becomes millivolts. However, the amplitude of EMG signals is in the range of
millivolts. Hence, when EMG signals are amplified more than 100 times, the ADC which follows
in the next stage should be given consideration of the DC bias threshold. Exceeding the threshold
will result in signal distortion. Accordingly, the LNA is required to have an adjustable gain with a
single external resistor that should not exceed 1 component per channel. This will aid the circuit
in the reduction of circuit complexity and additive noise. Additionally, the IC is expected to have
a wider range of gain so as to incorporate the smallest amplitudes

Moreover, component selection should consider overall circuit performance. One of the issues that
results in performance degradation is additive noise from components. Hence, for the purpose of
performance maximization, the LNA IC used must have a very low input referred noise so as to
avoid noise contribution.
28
4.2.2.2 CMRR
This is the most defining characteristics of the LNA of choice. This feature will be responsible for
the noise cancellation due to power line interference and noises from miscellaneous sources.
Accordingly, the amplifier is selected based on the level of CMRR.

4.2.2.3 Operating frequency

The signals handled by the circuit are low frequency signals. Hence, the LNA is expected to have
consistent performance in regards to gain and CMRR for the bandwidth of operation without any
glitches.

4.2.2.4 Power supply

The LNA is expected to operate within a wide range supply which allows the circuit design achieve
low power dissipation as well as capability of a single power supply to be used by other circuit
elements as well.

4.2.2.5 Release/revision dates

For the purpose of making this research relevant by paving options for future citations and
incorporation in related researches, the component should be up-to-date. Accordingly, for the LNA
to be considered in this work, two points are considered:

• Product availability on the market

• Cross checking for other products that offer superior performance without major variation
in cost, safety and availability

4.2.2.6 Application notes

The LNA is given higher priority if designated by the manufacturer that it has been specifically
designed for application in ECG, EEG, EMG and related low frequency & amplitude applications

In summary, using the research matrices mentioned earlier, extensive research through various IC
manufacturers is carried out. Even though the research in regards to manufacturers included
numerous leading IC vendors currently on the market, the dominant vendors available that provide
high-end products for medical instrumentation are Texas Instruments and Analog Devices. Some
of the ICs have been revised through the years through production. However, for the purpose of

29
research, all ICs are thoroughly reviewed and compared with the aid of the previously mentioned
comparison metrics. Table 5 summarizes the findings in accordance with the metric provided.

Accordingly, INA819 is chosen as it provides impeccable noise cancellation for the bandwidth of
operation. INA819 is a high-precision instrumentation amplifier that has flexibility in power
supply requirement and offers low input offset voltage and current noise. Moreover, this IC
satisfies one of the critical goals of this work by providing a high CMRR that exceeds 90dB at
minimum gain, G=1. The amplifier has specific applications in ECG amplifiers and general
medical instrumentation. A minified schematic of the LNA is shown in figure 4.1.

Figure 4.1 Minified internal schematic of INA819 [23]

The gain of the amplifier is set through the use of a single resistor between RG ports using the
equation given below [23].

50𝐾𝛺
G=1+ (4.1)
𝑅𝐺

30
 Table 4.2. Component survey for signal amplification
Minimum Maximum Noise at Supply Minimum Bandwidth Date of Specifically, Relevant
Gain Gain 1KHz Voltage Range CMRR (MHz) at Release/Revision Application Notes
(V/V) (V/V) (nV/√Hz) (Rail-to-Rail) (dB) Gmin
Part Number
Harsh environment data
AD8229 1 1000 1 8 - 34 80 15 February, 2012 acquisition
AD8237 1 1000 68 (+1.8) - (+5.5) 106 0.2 August, 2012 Medical Instrumentation
AD8420 1 1000 55 5.4 – 36 100 0.25 January, 2015 Medical Instrumentation
AD8421 1 1000 3.2 5 - 36 80 10 May, 2012 Medical Instrumentation
AD8422 1 1000 8 4.6 - 36 80 2.2 January, 2015 Medical Instrumentation
AD8429 1 1000 1 8 - 36 90 15 February, 2017 Medical Instrumentation
INA118 1 10000 10 2.7 - 36 107 0.8 January, 2018 Medical Instrumentation
INA128 1 10000 8 4.5 - 36 120 1.3 April, 2019 Medical Instrumentation
INA129 1 10000 8 4.5 - 36 120 1.3 April, 2019 Medical Instrumentation
INA141 10 100 8 4.5 - 36 117 1 April, 2019 Medical Instrumentation

INA333 1 1000 50 4.5 - 36 110 2 December, 2015 Medical Instrumentation

INA818 1 1000 8 4.5 - 36 140 2 April, 2019 Medical Instrumentation
INA819 1 1000 8 4.5 - 36 140 2 December, 2018 Medical Instrumentation
INA821 1 1000 7 4.5 - 36 140 4.7 December, 2018 Medical Instrumentation
INA828 1 1000 7 4.5 - 36 140 2 January, 2018 Medical Instrumentation
INA2141 10 100 8 4.5 - 36 117 1 August, 2018 Medical Instrumentation

31
4.2.3 Analog to digital conversion
Once the bio-potential signals have gone through the stage of amplification and conditioning, the
next step is the conversion of the analog signals to their digital equivalent. To do so, an ADC is
incumbent. The ADC is responsible to convert the analog signals to their digital representations
without distortion and send the converted data to a DSP through a certain communication protocol.
Some of the research angles applied for LNA selection is also applied for the survey and selection
of ADC such as input referred noise, CMRR, power supply requirement, release/revision dates
and application notes. Moreover, there are additional points that are specific for the ADC which
are discussed below.

4.2.3.1 Analog input channels

In regards to minimization of circuit complexity, the number of analog inputs available by the
ADC is very crucial. There are various types of ADCs currently available in the market that
provide different analog input channel capacity. However, as the circuit should be capable of
handling ECG, EMG and EEG signals, the total number of analog input channels required by the
ADC is 32. Hence, the choice of ADC is done with the total number required as per the total input
channels available kept in mind.

4.2.3.2 Sample rate

The IFCN standard (as given in table 4.1) requires the minimum sample rate to be 200.
Accordingly, the ADC of choice must meet this standard and provide a sample rate that is greater
than the minimum threshold. This sample rate must also satisfy the Nyquist criteria given in
equation 4.2 [26].

Fs = 2fmax (4.2)

4.2.3.3 Resolution
Similarly, the minimum number of bits required by the IFCN standard must be met by the ADC
of choice which refers to the resolution of the ADC. This will also contribute to the signal integrity
of the overall circuit.

32
4.2.3.4 Data rate and CMRR
Once the standards in regards to sampling and resolution are met, the CMRR performance against
variable data rate is also taken into consideration. This is done so as to ensure the flexibility of the
ADC for the purpose of scalability.

4.2.3.5 Conversion architecture

Of all the selection criteria mentioned earlier, the conversion architecture is the most critical in
regards to this research. This refers to the type of architecture the ADC implements for the
conversion of analog signals to their digital representations. ADCs can be categorized through
their conversion schemes: successive approximation, sigma-delta, flash converters, pipelined, and
bit-per-stage. This property becomes a defining one due to the direct implication it has on circuit
complexity as it comes with pre-requisites in regards to circuit components. For this work, an ADC
that employs oversampling principle is the most attractive since it drastically reduces the
requirement of anti-aliasing filters and hence significantly reduces circuit complexity as well as
noise.

In summary, inferring from the points discussed earlier, the ADC is selected through research from
manufacturers. Even though there are not ample choices available as in the case of LNA, there are
high precision converters available for the specific work at hand which is described in table 3.6.

Through a thorough analysis of features provided by various ADCs, the ADS1299 is selected for
the work at hand. The ADS1299 is a low-noise analog-to-digital converter specifically designed
for EEG and bio-potential measurements. One of the most unique features of this converter is the
fact that it allows to programmatically switch of analog input channel selection from multiplexed
to simultaneous sampling type and vice versa. It utilizes sigma-delta converter which drastically
eliminates the requirement of anti-aliasing filters that are required with typical Nyquist ADCs. The
converter has an on-board oscillator and programmable gain amplifiers. Moreover, the ADS1299
has been specifically designed for EEG and related bio-potential application which suits the scope
of this research.

33
 Table 4.3 Component survey of analog to digital converter

Number of Date of Relevant

Part Analog Input Conversion Programmabl Supply voltage Release / Application
Number Channels Architecture CMRR (dB) e Gain range Data Rate Revision Notes
Flash October, Medical
AD9249 16 N/A Unavailable 1.7 – 1.9 455MSPS 2013 imaging
Flash N/A January, Medical
AD9253 4 Unavailable 1.7 – 1.9 125MSPS 2018 ultrasound
Flash N/A April, Medical
AD9257 8 Unavailable 1.7 – 1.9 65MSPS 2013 imaging
Pipeline N/A April, MRI
AD9461 1 Unavailable 4.75 - 5.25 130MSPS 2006 receivers
delta-sigma EEG, ECG,
Available (1, 2, 250SPS - January, Sleep Study
ADS1299 8 110 4, 6, 8,16,24) 4.75 - 5.25 16kSPS 2017 Monitoring
delta-sigma Unipolar (0 - 5) Medical
Bipolar (-2.5 - March, Instrumentati
ADS1258 16 N/A Unavailable +2.5) 125kSPS 2011 on

34
4.2.4 System design review
Once components that fulfil the requirement of the designs are selected, a high-level design is
constructed representing the general operation mechanism of the signal acquisition system. The
design incorporates individual blocks the signals undergo from detection through to conversion
and transmission. The high-level block diagram of the acquisition hardware is shown in figure 4.2.
The diagram depicts the circuit blocks required in regards to the following key points:

• Safety regulations
• System-level performance requirements
• Signal-specific requirements

In addition to the critical components of the signal acquisition system being the LNA and ADC as
discussed in previous sections, the design must incorporate other fine details such as overpower
protection circuit, gain adjustment components, power supply requirement, communication
protocol etc.

4.3 Simulation
Implementation of the techniques discussed so far requires setting up a simulation model for bio-
potential signal acquisition with all non-idealities that occur in physical application of such
systems. The simulation model consists of two distinct stages: analog sub-system, and digital sub-
system.

In the analog sub-system, bio-potential signal models acquired from previous researches is used
as well as general signals that resemble the amplitude and frequency of such signals. The system
is simulated with an addition of noise for the purpose of imitating non-idealities that occur in actual
physical applications. The validation is done using the SPICE model provided by the manufacturer
with the aid of CAD tools.

In the digital sub-system, a Simulink model for ADS1299 is built with all the properties of the
converter put into consideration. The design variable is generate using SIMSIDES, a SIMULINK
tool, after which the simulation model is constructed.

35
Figure 4.2 High level block diagram
36
4.3.1 Bio-potential signal sources
If analyzed from a non-technical point of view, bio-potential signals, especially EEG signals,
resemble a noisy waveform. Hence, it is necessary to familiarize oneself with the waveforms of
the signal even though their properties in regards to amplitude and frequency have been mentioned
in previous section.

There are numerous models available to accurately model bio-potential signals. However, this
research uses Physionet.org, an online database that has a variety of actual records of bio-potential
signals of physical subjects made across the globe which are available in. edf format.

The data fetched from the database are converted to a signal source format compatible with PSpice
A/D and used for further evaluation. Summarized specification for implemented EEG and EMG
sources are given in tables 4.4 and 4.5

Table 4.4 Summary specification of EEG sample

EEG record Specification
Number of channels 23
Sampling rate 256Hz
Number of seizures in file 0
Total time 1 hour

Table 4.5 Summary specification of EMG sample

EMG record Specification
Number of channels 23
Sampling rate 4KHz
Sample profiles • Healthy
• Myopathy
• Neuropathy
Total time 1 hour

37
In simulating the bio-potential signal source, the two extremes selected are the EEG and EMG
signal. The justification used for this is the fact that EEG signals are the weakest amongst the types
of signals proposed for this work and EMG signals are the strongest in amplitude and frequency.
The time and frequency analysis of the signal sources are presented in Figures 4.3 – 4.6.

Figure 4.3 Time domain analysis of EEG source

Figure 4.4 Frequency domain analysis of EEG source

Figure 4.5 Time domain analysis of EMG source

38
 Figure 4.6 Frequency domain analysis of EMG source
4.3.2 System simulation models
As discussed in the previous section, the system design is broken down into two major parts as
analog sub-system and digital sub-system. In the analog sub-system, the major component is the
low noise amplifier selected for the work at hand whereas in the digital sub-section, the major
concern of design is the analog-to-digital converter. In order to assess the properties of the selected
component, the requirement is an analysis and simulation software to facilitate the Computer aided
Engineering (CAE). The CAE has two major elements which are: Computer aided design (CAD)
and computer aided manufacturing (CAM). For this specific work, the CAD handles issues related
to schematic design, simulation profile, PCB design and analysis. Accordingly, there are two CAE
tools utilized in this work: Cadence OrCAD V16.3 CAE software package and MATLAB R2018a.

4.3.2.1 Low noise amplifier

For accurately simulating the circuit behavior of INA819, Cadence OrCAD is used. The amplifier
is used to carry out two major tasks; Amplification and common mode signal rejection. In order to
simulate the performance of the component, the circuit configuration is set up as shown in figure
4.7. Through reference from the datasheet provided by the manufacturer, the CMRR is given as
110dB at an operating gain, G=10. However, the gain must be variable as the signals have different
amplitudes which will have an effect on the CMRR of the component. In order to test the
performance of the signal, a common mode signal of variable amplitude is supplied to the circuit
and evaluated at different levels of voltage gain. The variables of the simulation profile are
provided in table 4.6.

39
As provided in table 4.6, the simulation profile takes into account two major elements: Amplitude
and frequency variations of ECG, EMG and EEG signals; 50/60 Hz power line interference.
Accordingly, the output in regards to variable input is recorded and performance analysis is carried
out. Moreover, the CMRR is evaluated with the circuit configuration provided in figure 4.7 using
equation 4.3 [25].

Vdiff
CMRR = 20log10 (4.3)
Vcm

Figure 4.7 Circuit configuration of INA819 [23]

The performance analysis is carried out through amplitude and frequency variation of differential
signal input in regards to common mode signal input. Table 4.6 summarizes the input variation of
the simulation profile.

40
 Table 4.6 Simulation profile of INA819
Quantity Variation
Vcm (mV) 3 –100
Vdiff (mV) 1 – 1000
Fcm (Hz) 50 – 60
Fdiff (Hz) 1 – 200,000
Rg 50 – 50,000
Bias (±9) – (±18)

4.3.2.2 Sigma-delta ADC

The other defining component of this research is the analog-to-digital converter. Through the
comparison chart provided in table 4.3, the ADS1299 is an impeccable choice when it comes to
conversion of bio-potential signals of very weak amplitude to their digital counterparts. In order
to accurately carry out the simulation of the ADC, a detailed study of previous performance
analysis documents as well as the device property is carried out. In doing so, a simulation model
of the converter is built using MATLAB R2018a and SIMULINK with the aid of DSToolbox. As
shown in figure 4.8, the ADC is comprised of distinctive stages that have diverse functions.
However, the critical module that is considered as the defining characteristics of the converter is
the ΔΣ modulator. This modulator is responsible for oversampling the input signals and passing it
to the control interface for communication and further processing. The fact that the conversion
architecture is a ΔΣ has large contribution in addressing the anti-aliasing filter design requirement,
speed as well as general circuit complexity. A minified depiction of the operation mechanism of
the ADC is shown in figure 4.2.

The simulation model for assessing the requirement conformity of the device is accurately
constructed using a MATLAB – SIMULINK co-simulation. MATLAB is used to generate the
modulator parameters with the aid of DSToolbox through a multi-step process. Once the
parameters are successfully generated, SIMULINK is used to put generated parameters into the
simulation with the addition of non-idealities that mimic practical applications. Moreover, the
digital/decimation filter is constructed providing a simulation model to test the performance of the
system altogether.
41
 Figure 4.8 Block level analysis of ADS1299
When designing a simulation model for ADS1299 there are numerous determining specifications
that are considered. Major design specifications are made through component evaluation which
are described in table 4.7.

Table 4.7 Simulation specification for ADS1299

Property Specification
Modulator
Order 2nd
Fs 1.024MHz
OSR 2560
Fb 200Hz
Architecture Feedforward
Digital Filter
Type Sync
Order 3rd
Decimation Ratio 64

42
Once the specifications of the modulator in regards to order, OSR and modulator architecture are
identified, DSToolbox is used to generate the parameters and finally add non-idealities to mimic
real life applications. The process for system specification generation in MATLAB follows:

1. Determination of modulator order, OSR, quantizer level and modulator type

2. NTF synthesis
3. Realization of [a,b,c,g] coefficient of the modulator
4. Coefficient mapping to internal states ABCD and scaling
5. Realization of coefficient of [a,b,c,g] by ABCD with manual round-off

Once the modulator coefficients are successfully generated, the coefficients are exported to a
SIMULINK model which is constructed with non-idealities to mimic practical application. The
non-idealities considered in the model are:

• Clock jitter
• Switch non-linearity
• Thermal noise
• Op-amp noise

The simulation output is compared in regards to SNR output for ideal and non-ideal situations.
The non-ideal output is taken from the simulation output whereas the ideal is calculated through
initial specifications given in table 4.2 using the equation 4.4 [27].

3𝜋 𝑂𝑆𝑅 2𝑁+1
SNR =( 2 ) (2𝑁 + 1)( ) (4.4)
𝜋

4.4 Hardware Design

One of the metrics of this research is to come up with a portable hardware that can be inter-
changeably used for the detection of ECG, EEG and EMG signals. Accordingly, Cadence OrCAD
v16.6 CAE software package is used in regards to circuit schematic, PCB layout, routing and
simulation. More specifically, Cadence OrCAD Schematic capture is used for the schematic design
and Allegro Expert for the PCB layout, routing and simulation.

The system design can be categorized in the mixed signal application genre as it contains both
analog and digital sections. Basically, a PCB design and assembly consist of the bare board,

43
attached components and connectors which altogether have a significant impact on the process of
component placement, product reliability and troubleshoot. Hence, in order to make the hardware
meet these specifications, the hardware design is done with two distinct considerations which are:
electrical considerations; mechanical and thermal considerations.

However, before diving into the PCB design, all circuit elements with their specifications are
clearly defined as per their respective datasheets and required functionality. Moreover, their layout
in regards to electrical connectivity is done through individual schematic projects where the blocks
are: overvoltage protection; signal amplification; analog-to-digital conversion; digital signal
processing. The circuit schematic of all the major blocks are described below.

4.4.1 Overvoltage protection

When comparing the hardware for ECG, EEG and EMG applications, the major concern is against
defibrillation which refers to the stimulation of the heart through the use of electrical shock for the
purpose of resetting the electrical state and adjust the rhythm [28]. There are numerous approaches
one can implement for the protection circuitry. In this research, a neon lamp with large voltage
rating along with Zener diodes is used. This configuration is used through comparison of available
methods in medical instrumentation guides. The circuit schematic for a single electrode is shown
in figure 4.9.

Figure 4.9 Overvoltage protection circuit

The circuit shown in figure 4.9 will only be used for selected pins under ECG applications. The
amplitude of applied voltage in defibrillation differs under variable conditions. However, the
amplitude can reach up to 5kV in clinical application. However, for EEG and EMG applications,
the internal overvoltage protection of the INA819 of 60v is assumed to be sufficient enough.
44
4.4.2 Signal amplification
As discussed in previous section, the INA819 is used for the amplification of the bio-potential
signals and achieve a high level of CMRR. The electrical schematic for connecting the electrodes
to the amplifier for a single channel is shown in figure 4.10.

Figure 4.10 INA819 circuit schematic

The major consideration in the signal amplification signal is the difference of reference points for
different measurement types. For EEG and EMG measurement, the positive input of the amplifier
is taken from different measurement points according to the measurement standards mentioned
earlier whereas the negative input is taken from the scalp of the patient for EEG and elbow for
EMG. However, for ECG signal amplification, the 12-lead system with the derivation provided in
table 1.3 is used in which the signal designated by Vwct is a Wilson central: Signal average of
LA (left arm), RA (right arm) and LL (left leg).

Figure 4.11 Wilson central circuit for referencing voltages of unipolar ECG
45
As can be inferred from table 1.3, the measurements for lead III, aVR, aVL and aVF are derived
from other measurements. Accordingly, this work proposes for the measurement of these signals
to be done in the DSP which eliminates the necessity for building circuit for the manipulation of
these signals.

4.4.3 Analog to digital conversion

Once signals have gone through the amplification process, the next step is the conversion of the
analog signals to a digital form. It has been discussed in the previous section that the ADS1299
analog frontend is proposed for this purpose due to its exceptional performance. However,
ADS1299 has 8 inputs whereas the application requires 32 analog channels. Therefore, there must
be 4 ADS1299 devices in place to meet the desired specifications.

According to the manufacturer’s datasheet, there are two modes to choose from when using more
than one device in a single application [24]: cascaded mode; daisy-chain mode. The choice will
have a direct impact on the required number of chip-select lines. For this work, the devices are
connected in a daisy-chain mode which requires only a single chip-select line. The functional block
diagram of the operation is depicted in fig 4.12. However, when operating multiple devices using
a daisy-chain mode, the maximum number of devices allowed depends on the data rate at which
the devices are operated determined by equation 4.5 [24].

fsclk
Ndevices = (fDR)( Nchannels)(Nbits)+ 24
(4.5)

where Nbits = device resolution, and

Nchannels = number of channels on the device
fsclk = frequency of serial clock
fDR = data rate

46
Hence, using equation 4.5, the minimum required SCLK speed to operate 4 devices with 16kSPS
in daisy-chain mode is 3.1MHz. A functional block diagram of the devices connected in daisy-
chain mode is shown in figure 4.12.

Figure 4.12 Daisy chain configuration for ADS1299

4.4.4. PCB design

The PCB design is inclusive of numerous processes such as electrical schematic, layout and
routing. Main considerations taken in regards to the design is layer stack-up, trace width and trace
spacing.

47
The final design of the circuit contains a mixture of both analog and digital circuit components
with their respective power requirement and grounding options. Accordingly, the layer stack-up
definition is done so as to separate analog and digital components in regards to their power
requirements, routing, as well as placement. Critical circuit components along with their
specifications is given in table 4.8.

Table 4.8 Power and grounding requirement of core components

Component Power requirement Grounding requirement
INA819 Analog Analog
ADS1299 Mixed Mixed
TMS320C5514 Mixed Mixed
TL6012 Analog Analog
LTC3261 Analog Analog
LM3480 Analog Analog

Taking into consideration the type of power and ground required by circuit components, the layer
stack-up is defined by taking scalability, placement, routing and spacing into consideration. The
PCB is completed in 10 layers whose name and function are provided in table 4.9.

Table 4.9 Summary of layers with functions

No. Layer name Layer type
1 TOP Routing
2 V+ Plane
3 AGND Plane
4 V- Plane
5 ABOT Routing
6 SHLD Plane
7 DVDD Plane
8 DGND Plane
9 AVDD Plane
10 BOTTOM Routing
48
Another crucial property that requires careful analysis is the trace width and spacing used in the
design. As can be inferred from table 4.9, the circuit has 6 global power nets which are:

• Analog power (V+, V-) for the circuit elements such as INA819 and LDO regulators;
• Analog and digital grounds;
• Digital and analog supply for digital circuit components of ADS1299 and TMS320C5515.

For the determination of trace width requirement, there are two considerations made for the circuit
components: Current handling capability and Impedance. Through a careful study of individual
datasheet for each component, the short circuit output current for the devices is determined after
individual traces are calculate using equation 4.5 [29].

1 𝐼
𝑤 = ( 1.4 ∗ ℎ ) ( 𝑘 ∗ ∆𝑇 0.421 )1.379 (4.5)

w is the minimum trace width, h is the thickness of the copper cladding, 𝐼 is the current load of the
trace, k = 0.024 for inner layers and 0.048 for outer layers. Hence, the tracing requirement is
different from one component to the other whilst being dependent on whether the top/bottom
routing layers or the inner routing layer being used.

The other critical component taken into consideration in regards to signal integrity of the circuit is
the trace spacing requirement which is dependent on the peak-to-peak voltage of components. This
property is very crucial as there is a high density of instrumentational amplifiers used in the system
which gives rise for the overall crosstalk. Hence, accurately determining the minimum allowed
space between traces will aid the minimization of crosstalk. Once the minimum value is
determined, the 3w rule is implemented for the purpose of signal integrity. The standard used is
summarized in table 4.10.

Table 4.10 Minimum trace spacing

VDC or Vp-p Internal traces (Mils) External traces (Mils)
0 – 15 2 4
16 – 30 2 4
31 – 50 4 24
51 - 100 4 24

49
Table 4.10 is used to determine trace spacing for the entirety of the PCB design. A 3D view of
the finished layout is given in the figure below.

Figure 4.13 3D Layout of the PCB design

4.5 Simulation results and discussion
4.5.1 Signal amplification
The first stage of the circuit is used to amplify low amplitude signals whilst rejecting common
mode noise. Accordingly, the simulation is carried out to test these functionalities of the INA819
with a gain of, G = 500. Once amplified, the output data is studied in regards to its amplitude as
well as noise artifact.

Bio-potential signals from a multichannel setup are passed through INA819 operating in
differential mode. A common gain of 500 is used for all signals through an external resistor of
100Ω using the configuration given in figure 4.13. Input-output relation is shown in figures below.

50
 Figure 4.14 Bio-potential signal input

Figure 4.15 Bio-potential signal output

4.5.2 Effects of noise
The amplifier is used to amplify as well as reject common mode noise that are added as artifacts
to useful frequency component of the signals. For accurately simulating the effects of noise on the
bio-potential signals handled by the system, noise signals are integrated to the bio-potential signal
generators. The generated noise signals are meant to cover the wide range of noise interference
that occur in physical application. The noises with which the amplifiers CMMR is evaluated
against are power line interference noise and white noise [38].

Analysis of power line interference noise is done using a sinusoidal signal of 50Hz frequency and
3mVpp amplitude. On the other hand, white noise meant to account for the miscellaneous noises
with variable spectral presence that occur in physical applications is added to the bio-potential
signal. These noises include interferences that arise from miscellaneous operating conditions such

51
as electrical machinery, fluorescent lighting, computer and peripheral accessories, thermal noise
etc.

Accordingly, from the three types of bio-potential signals, EEG signals possess the weakest feature
in regards to both amplitude and frequency. Hence, the effect of noise is simulated on the EEG
signal. Miscellaneous noises mentioned earlier are integrated to the EEG signal generator. Time
and frequency domain analysis of the noisy input and achieved output are given in the figures
below.

Figure 4.16 Time domain analysis of power line interference noise

52
 Figure 4.17 Frequency domain analysis of power line interference noise

Figure 4.18 Time domain analysis of generated white noise

Figure 4.19 Frequency domain analysis of generated white noise

The generated noise signals are added to the bio-signal signal source. The totality is a bio-potential
signal that possesses both wanted and unwanted frequency components. This signal is fed into the
amplifier operating in differential mode. The input signal with additive noise components and
amplified output with common mode signals rejected in its time and frequency domain analysis is
shown in the figures 4.19 – 4.22.

53
Figure 4.20 Time domain analysis of signal input with added noise signals

Figure 4.21 Frequency domain analysis of input with added noise signals

54
 Figure 4.22 Time domain analysis of amplified output

Figure 4.23 Frequency domain analysis of amplified output

4.5.3 CMRR analysis
Once the noise cancellation performance of the amplifier is analyzed, the CMRR stability is
verified. As one of the metrics of this work is achieving stable CMRR for the bandwidth of the
bio-potentials operated by the system, the CMRR for the bandwidth of application is simulated
using the same circuit setup in figure 4.7 and analyzed for variable gain conditions whilst applying
common mode noise interferences. The CMRR performance for a wide frequency is shown in
figure 4.23.

Figure 4.24 CMRR plot of distinct gain levels

4.5.4 PSRR analysis
A common performance analysis metric for electronic circuit design is the power supply rejection
ratio. The simulation is done up to a maximum of 100 KHz frequency with a source of ±15V with

55
1Vac for carrying out transient analysis on the circuit. The amplifier is tested for power supply
ripple effects at different gain and the result is shown in figure 4.24.

Figure 4.25 PSRR plot for distinct gain levels

4.5.5 Analog to digital conversion
The section for ADC has two critical processes which are oversampling and decimation. Initially,
the input signal is oversampled at a given oversampling ratio. Finally, the second stage introduces
a tradeoff between signal resolution and data rate by decimating the signal at a certain rate. The
simulation profiles used to test the ADC block is taken from table 4.7 with added non linearity
described in section 4.3.2.2. The DSToolbox SNR analysis gives an estimate of the ratio to be
156.6dB. However, the signal output in the actual simulation setting with added real-time factors
such as op-amp noise, clock jitter and more has yielded an SNR of 113dB. The serial bit stream
along with the reconstructed data is shown in the figures 4.25 and 4.26 respectively.

Figure 4.26 Snap shot of serial data captured through a serial monitor

56
 Figure 4.27 Reconstructed serial stream
4.4.6 Performance analysis
The benchmark for performance evaluation and comparison is done in regards to an existing BCI
hardware as well as researches published within the past years. Finally, the design is checked
against compliance conformity with the IFCN standard.

To analyze the proposed design in regards to the points mentioned above, the open source cyton
board by OpenBCI is taken as the bench mark for available products. The cyton board is a hardware
manufactured and maintained by researchers in the field of bio-medical research and development.
The board has 8 bio potential input channels powered by ADS1299 as the critical circuit
component. It has an inbuilt SD storage and wireless communication. However, the board cannot
be used for actual clinical purposes. However, there are numerous researches available that are
based on this board which mainly span the areas of EEG and EMG applications. A summary of
the features of the cyton board is given in table 4.11.

Table 4.11 Feature summary of cyton board

Feature Value
CMRR (dB) 120
LPF (Hz) 100
Sample rate 500
Bits 24
Channel 8
Main application EEG signal acquisition

57
Even though the cyton board cannot be used for clinical purposes, it can be scaled and adjusted for
clinical application. Accordingly, there are numerous researches published that are based on the
cyton board that are diverse in features as well as application areas. However, the major motive
for researchers in the area is not the board by itself. Rather, the critical point of attraction is the
ADS1299 SOC. As the SOC is specifically designed for EEG and ECG applications, it has been
the focal point of these researches. Table 4.12 summarizes features of sample researches that are
based on ADS1299 SOC.

Table 4.12 Summary of ADS1299 based publications

Year of Application Number of Resolution Sample rate
publication Channels
2019 [31] Brain-to-Brain interface 8 24bits 250Hz
2019 [32] BCI 16 24bits 125Hz
2019 [33] sEMG 4 24bits 250Hz
2017 [34] EMG/EOG 4 24bits 250Hz
2017 [35] BCI - IOT 8 24bits 250Hz
2016 [36] Automated brain switch 8 24bits 1KHz
2016 [37] Robotics 8 24bits 250Hz

Ching-sung wang [14] implemented a thorough design of a 32-channel EEG system for BCI
application which follows a general architecture similar to the one described in figure 1.5. The
study integrates a mixed signal design and development that supports a software interface to
achieve the design goal. For the purpose of noise cancellation, a preamplifier with high CMMR
and SNR is integrated and tested. Moreover, the design includes adjustable amplification and filter
function meant to be used for different EEG frequency bands. The final design is confirmed to
meet IFCN standards and measurement verification conducted to calibrate the accuracy and
reliability of the system. The summarized list of features of the output is provided in table 4.13.

58
 Table 4.13 Ching-sung wangs’ 32 Channel EEG acquisition system design feature summary
Feature Value
CMRR (dB) 130
HPF (Hz) 0.16
LPF (Hz) 100
Sample rate 500
Bits 16
Channel 32
Main application EEG signal acquisition

4.4.6.1 Scope of application

The goal of this research is to design a multipurpose hardware for bio potential signal acquisition.
Hence, the portable device should be capable of implementation in EEG, EMG and ECG signal
acquisition. When taking a thorough review of the researches and products discussed in the
previous section, it is evident that most designs are done for applications in the areas of EEG and
sEMG data acquisition. Accordingly, there is a clear absence of ECG specific design
functionalities such as a body reference montage i.e. Wilson central. Moreover, most researches
aren’t meant for clinical application with a few exceptions. On the contrary, this work proposes
the hardware to be used for diagnostic purposes in actual clinical applications.

4.4.6.2 Channel capacity

With the exception of Ching-sung wang [14], the researches summarized in table 4.12 provide a
maximum of 8 number of analog input channel. However, this design proposed in this research
provides a 32 analog input channel which can be utilized for EEG, EMG and ECG applications.

4.4.6.3 Noise reduction

Achieving a high level of CMRR is the focal design parameter in this work. The fact that INA819
is a product released in 2018, there has not been many designs that utilize its high level of noise
rejection and medical instrumentation application. Accordingly, the implementation of the LNA
in this work has yielded a high CMRR of 145dB for a wide range of frequency application with
stability. All the researches summarized in table 4.12 utilize the intrinsic CMRR capability of

59
ADS1299 has a typical value of -110dB. However, the integration of INA819 as the front end
LNA in this work has allowed the analog input to achieve a stable CMRR of 145dB for the
frequency band of application. In comparison, this research yields a 21% improvement from
Ching-S

4.4.6.4 Analog to digital conversion

The researches summarized in table 4.12 utilize the ADS1299. However, since the maximum
number of channels in the researches is limited to 8, there is only a single ADS1299 used in all the
designs. However, this work proposes a 32-channel simultaneous sampling montage for the
ADS1299 through the use of 4 independent ADS1299 SOCs operating in daisy chain mode whose
advantage is twofold. Firstly, the implementation of simultaneous sampling for all 32 available
channels provides an ease in regards to visibility for GUI applications. And secondly, using a
daisy-chain module rather than cascaded mode has led to a reduced trace runs in the case of the
PCB design.

4.4.6.5 Standard compliancy

Even though it cannot be said that the design in this work is in compliant with every standard that
exist for bio-medical application development, the initial metric of IFCN standard compliancy is
achieved. The compliancy check of the design proposed in this work against major IFCN standard
is described in table 4.14.

Table 4.14 Comparison of proposed design with IFCN standards

IFCN This research
CMRR ≥110dB 145dB
Rin ≥100MΩ 100GΩ
Sample rate ≥200 16
Resolution ≥12 bits 24 bits
Channel ≥24 32

60
 4.4.6.7 Summary
The overall performance analysis is done on both spectrum of the research through product as well
as previous works in the area. Accordingly, the analysis shows that the design of the signal
acquisition kit possesses enhanced features in regards to channel capacity, noise performance and
overall issues of circuit complexity. Table 4.15 presents a summary of the comparison done in
regards to both product and research.

Table 4.15 Summary of comparison in product, research and standard

Open BCI Ching-sung Wang [14] This research
CMRR (dB) 110 130 145
Sample rate 1.024M 500 1.024M
ADC Resolution 24 24 24
Channels 8 32 32
Circuit complexity Highly optimized Less optimized Better optimized
Scope of application EEG Mainly EEG but suggest ECG usage EEG/ECG/EMG
as well
IFCN compliancy Does not comply Complies Complies

Analyzing table 4.15, the comparison yields a 21% improvement in common mode rejection ratio
from previous researches and existing BCI products. Moreover, the analog input channel capacity
is boosted by 33% in comparison to the cyton board of Open BCI. Additionally, the diagnostic
capability of the data acquisition system is enhanced due to the capability of utilizing EEG, ECG
and EMG detection on the same device.

61
 Chapter 5
Conclusion and future work
5.1 Conclusion
The front-end design proposed in this research presents a generic approach for designing bio-
potential signal acquisition devices that can be implemented as a multi-purpose portable hardware.
The design takes into account research of recently manufactured ICs and SOCs specifically
designed for the purposes of medical instrumentation. Aside from the specific objectives, a generic
simulation model for a sigma delta modulator with non-idealities is availed for future inferences.

In this research, a detailed research and simulation-based verification for low noise amplifiers is
done with the major criteria being high common mode rejection ratio with stability over a wide
range of frequency application. The amplification capability along with noise cancellation and
common mode rejection stability is verified through the use of actual bio potential signals acquired
from international bio physical signal databases. The overall performance is evaluated in ideal
scenarios with the absence of noise signals as well as through the application of power line
interference and white noise. The critical feature of the digital section of the design is considered
to be the sigma delta modulation ADC with simultaneous simulation. For the purpose of design
and validation, a standard simulation model of an analog to digital converter with sigma delta
modulation architecture is built and tested in ideal and non-ideal scenarios. The hardware is
designed for the purpose of implementing a 32-channel bio-potential signal detection through
careful consideration of electrical and mechanical constraints.

The obtained simulation results are compared with selected researches as well as commercially
available products in regards to application areas, noise cancellation and digital design
performance metrics. The evaluation of the implemented low noise amplifier yielded an average
of 21% improvement in common mode rejection ratio in comparison to previous researches and
existing BCI products. A 33% improvement in analog input channel capacity is achieved whilst
broadening the application to a multi-purpose portable hardware capable of being used for EEG,
EMG and EEG application. Moreover, standard compliancy in regards to the IFCN standard is
carried out.

62
5.2 Limitations
The design proposed in this work has several limitations which must be taken into account when
utilizing the results as well as design. Although the simulation models used in this work are highly
accurate and well maintained by the manufacturers, it doesn’t account for the fact that experimental
verification has not been done. The main reason for this is the availability of the products on the
local market and allocated budget. As the thesis work is done through self-sponsorship, funding
has been a very critical drawback in verifying the design experimentally.

The other core limitation is that the design might be considered as lacking in regards to circuit
optimization as it uses a large number of low noise amplifiers which can also be considered as
redundant since the ADS1299 also provides excellent noise cancellation scheme. However, the
fact that the implemented low noise amplifier offers a significant improvement in noise
cancellation makes up for the increased size of hardware. Moreover, noise cancellation can be
improved through the use of a right-leg drive circuit which is available on the existing component
but left unused due to the choice of design.

5.3 Future work

The ultimate objective of this work is to design a multi-purpose signal acquisition device that can
be used for EEG, EMG and ECG applications. The design can be further optimized in regards to
complexity and cost in order to simplify the medical provision for people in low-income countries.
The hardware kit can be distributed to health centers across the countries which will have a
significant contribution in cutting-back time and money spent by patients and institution.

Moreover, the digital design can be further extended to have a data syncing protocol to a remote
repository (database) that allows patient’s data review by medical professionals that reside outside
of the patient’s locale. This is very helpful fin scenarios where a professional opinion is required
for further scans and treatment which created a foundation for online collaborative diagnosis.
Furthermore, the online repository can be used for the purposes of statistical data generation.

63
 References
[1] W. Van Drongelen, “Signal Processing for Neuroscientists”. San Diego: Elsevier Science &
Technology Books, 2010.
[2] C. W. John, “Medical Instrumentation: Application and Design (edited by john G.
Webster)”, Fourth edition, John Wiley and Sons Ltd, 2010.
[3] J. Enderle and J. Bronzino, “Introduction to biomedical engineering”. Amsterdam:
Elsevier/Academic Press, 2012.
[4] S. Sanei and J. Chambers., “EEG Signal Processing”. John Wiley & Sons, 2011.
[5] E. Ashley and J. Niebauer, “Cardiology explained”. London: Remedica, 2004.
[6] “10-20 system (EEG)”, En.wikipedia.org, 2019. [Online]. Available:
https://en.wikipedia.org/wiki/10%E2%80%9320_system_(EEG) [Accessed: 06- Aug-
2019]
[7] F. Port., "Finding Power Spectrum of EMG Signal Acquired through Serial
Port.", Forums.ni.com, 2020. [Online]. Available: https://forums.ni.com/t5/Biomedical-
User-Group/Finding-Power-Spectrum-of-EMG-Signal-Aquired-through-Serial-Port/td-
p/3443060?profile.language=en. [Accessed: 03- Jul- 2019].
[8] A. Ra, U. Mahadev and M. Gaikwad, “Design of ECG instrumentation and
implementation of digital filter for noise reduction”, 2013
[9] “Clinical Neurophysiology”. [Online]. Available:
https://www.journals.elsevier.com/clinical-neurophysiology/view-for-free/guidelines-
of-the-ifcn-2nd-ed-published-1999. [Accessed: 17- Feb- 2019].
[10] "Muscle Control V1.1 — Open Bionics", Open Bionics, 2020. [Online]. Available:
https://openbionicslabs.com/obtutorials/artichoke-muscle-control. [Accessed: 06- May-
2019].
[11] J. Malmivuo and R. Plonsey, “Bioelectromagnetism”. New York: Oxford University Press,
1995
[12] N. B. Robert, “Analysis and Application of Analog Electronic to Biomedical Application”,
CRC Press LLC, 2003.

64
[13] D. Prutchi, M. Norris, “Design and Development of Medical Electronic Instrumentation”,
First edition, John Wiley and Sons Ltd, 2005.
[14] C. Wang, “Design of a 32-Channel EEG System for Brain Control Interface Applications”,
2012
[15] J. G. Webster, “Encyclopedia of Medical Devices and Instrumentation”, Second edition,
John Wiley and Sons Ltd, 2006
[16] Sarmiento, J. F.; Bastos, T. F.; Botti, A. B.; Elias, A.; Fri16z[]era, A.; Hubner, M.; Silva, I. V.
Characterization and diagnosis of fibromyalgia based on fatigue analysis with sEMG
signals. In 2012 ISSNIP Bio signals and Bio robotics Conference: Bio signals and Robotics
for Better and Safer Living (BRC); 2012; pp. 1–4.
[17] P. Palmes, W. T. Ang, F. Widjaja, L. C. Tan, W. L. , “Pattern Mining of Multichannel
sEMG for Tremor Classification”. IEEE Trans. Biomed. Eng. 2010, 57, 2795–2805,
doi:10.1109/TBME.2010.2076810.
[18] Gutiérrez, S. J.; Cardiel, E.; Hernández, P. R. A muscle fatigue monitor based on the
surface Electromyography signals and frequency analysis. In 2016 Global Medical
Engineering Physics Exchanges/Pan American Health Care Exchanges (GMEPE/PAHCE);
2016; pp. 1–6.
[19] N. Hazarika, A. C. Tsoi, and A. A. Sergejew, “Nonlinear considerations in EEG signal
classification,” IEEE Transactions on Signal Processing, vol. 45, pp. 829–836, April 1997.
[20] H.-H. Chang, J. M. F. Moura, Y. L. Wu, and C. Ho, “Automatic detection of regional heart
rejection in USPIO-enhanced MRI,” to appear IEEE Transactions on Medical Imaging.
[21] B. S. Carter, T. H. Beaty, G. D. Steinberg, B. Childs, and P. C. Walsh, “Mendelian
inheritance of familial prostate cancer,” Proceedings of the National Academy of
Sciences of the United States of America, vol. 89, pp. 3367–3371, April 1992.
[22] L. Frenzel, Electronics explained. Kidlington, Oxford, United Kingdom: Newnes, 2018, pp.
291-324.
[23] Texas Instruments, “Low-Power Precision Amplifier”, INA819 datasheet, Dec. 2018
[24] Texas Instruments, “Analog-to-Digital Converter for EEG and Biopotential
Measurements”, ADS1299 datasheet, July 2012 [Revised Jan. 2017]
65
[25] “Common-Mode Rejection Ratio (CMRR) – NI Digital Multimeters Documentation”,
Documentation help, 2020 [Online]. Available: https://documentation.help/NI-Digital-
Multimeters/cmrr.html. [Accessed: 16- Jan-2019]
[26] “Nyquist-Shannon sampling theorem”, En.wikipedia.org, 2019. [Online]. Available:
https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem.
[Accessed: 21- Jan-2019]
[27] X. Xing, P. Zhu and G. Gielen, Design of power-efficient highly digital analog-to-digital
converters for next-generation wireless communication systems. Springer, pp. 40 – 41,
2018.
[28] “Defibrillation”, En.wikipedia.org, 2019. [Online]. Available:
https://en.wikipedia.org/wiki/Defibrillation. [Accessed: 14- Oct-2019]
[29] K. Mitzner, B. Doe, A. Akulin, A.Suponin and D. Muller, “Complete PCB Design Using
OrCAD Capture and PCB Editor”, 2nd Edition, 2018
[30] S. Chen and Y. Chen, "Hardware Design and Implementation of a Wavelet De-Noising
Procedure for Medical Signal Preprocessing", Sensors, vol. 15, no. 10, pp. 26396-26414,
2015
[31] Jiang, L., Stocco, A., Losey, D.M. et al. BrainNet: A Multi-Person Brain-to-Brain Interface
for Direct Collaboration Between Brains. Sci Rep 9, 6115 (2019).
https://doi.org/10.1038/s41598-019-41895-7
[32] Julia, M., Ryan, P., Stephanie L, et al. “Embodiment improves performance on an
immersive brain computer interface in head-mounted virtual reality”. (2019).
https://doi.org/10.1101/578682
[33] Cao, J. , Tian, Z. and Wang, Z., “Hand Gestures Recognition Based on One-Channel
Surface EMG Signal”. Journal of Software Engineering and Applications, 12, 383-392.
(2019) doi: 10.4236/jsea.2019.129023.
[34] Micholaj, B. and Marcin, J., “Project and evaluation of EMG/EOG Human-Computer
interface”. Article in Przeglad Elektrotechniczny, 93, 128-130. (2017)
doi: 10.4236/jsea.2019.129023.

66
[35] Aries, T., Mauricio, C., Marcus, E. et al. “Using Brain-Computer Interface and Internet of
Things to Improve Healthcare for Wheelchair Users”. The Eleventh International
Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies (2017)
[36] R. V. Sastry, R. Aravind, G. Panwar, S. Indrapriyadarsini and A. G. Ramakrishnan,
"Automatic user customized brain switch," IEEE Annual India Conference (INDICON),
Bangalore, 2016, pp. 1-5.
[37] C. J. Perera, I. Naotunna, C. Sadaruwan, R. A. R. C. Gopura and T. D. Lalitharatne, "SSVEP
based BMI for a meal assistance robot," IEEE International Conference on Systems, Man,
and Cybernetics (SMC), Budapest, 2016, pp. 002295-002300.
[38] A. Velayudhan and S. Peter, "Noise Analysis and Different Denoising Techniques of ECG
Signal – A Survey, International Conference on Emerging Trends in Engineering &
Management", 2016, pp. 40-44.
[39] Neuman R. Michael, "Medical Instrumentation: Application and Design (edited by john
G.Webster) ", Fourth edition, John Wiley and Sons Ltd, 2010.
[40] J.R. Wolpaw, N. Birbaumer, D.J. McFarland, G. Pfurtscheller, and T.M. Vaughan, "Brain-
computer interfaces for communication and control". Clinical Neurophysiology, 113, 6,
767–791, (2002).
[41] M.A. Lebedev and M.A.L. Nicolelis, "Brain-machine interfaces: past, present and future".
Trends in Neuroscience, 29(9), 536–546, (2006).
[42] H. Chang, J. M. F. Moura, “Biomedical Signal Processing,”, ed. Myer Kutz, in Biomedical
Engineering and Design Handbook, 2nd Edition, Volume 1, McGraw Hill. 2010, pp. 559-
579.

67
 Appendix
Circuit schematics
Power supply

Electrode wiring

68
Amplifier array sampole

ADC wiring for 8-channel input

69
DSP connection setup

N.B. The schematic model is built using two individual blocks for the sake of clarity and doesn’t represents two devices.

70