Data acquisition system of 16-channel EEG

based on ATSAM3X8E ARM Cortex-M3 32-bit

microcontroller and ADS1299
Cite as: AIP Conference Proceedings 1862, 030149 (2017); https://doi.org/10.1063/1.4991253
Published Online: 10 July 2017

L. O. H. Z. Toresano, S. K. Wijaya, Prawito, et al.

ARTICLES YOU MAY BE INTERESTED IN

EEG data acquisition system 32 channels with relative power ratio based on Raspberry Pi 3
AIP Conference Proceedings 2168, 020017 (2019); https://doi.org/10.1063/1.5132444

Data acquisition instrument for EEG based on embedded system

AIP Conference Proceedings 1817, 040009 (2017); https://doi.org/10.1063/1.4976794

EEG data acquisition system 32 channels based on Raspberry Pi with relative power ratio and
brain symmetry index features
AIP Conference Proceedings 2193, 050003 (2019); https://doi.org/10.1063/1.5139376

AIP Conference Proceedings 1862, 030149 (2017); https://doi.org/10.1063/1.4991253 1862, 030149

© 2017 Author(s).
 Data Acquisition System of 16-channel EEG
Based on ATSAM3X8E ARM Cortex-M3
32-bit Microcontroller and ADS1299
L. O. H. Z. Toresano1, S. K. Wijaya1, a), Prawito1, A. Sudarmaji1, and C. Badri2
1
Department of Physics, Faculty of Mathematics and Natural Sciences (FMIPA),
Universitas Indonesia, Depok 16424, Indonesia
2
Biomedical Technology-TBM, Postgraduate Program, Universitas Indonesia,
Kampus UI Salemba, Jakarta 10430, Indonesia
a)
Corresponding author: [email protected]

Abstract. The prototype of the EEG (electroencephalogram) instrumentation systems has been developed based on 32-
bit microcontrollers of Cortex-M3 ATSAM3X8E and Analog Front-End (AFE) ADS1299 (Texas Instruments, USA),
and also consists of 16-channel dry-electrodes in the form of EEG head-caps. The ADS1299-AFE has been designed in a
double-layer format PCB (Print Circuit Board) with daisy-chain configuration. The communication protocol of the
prototype was based on SPI (Serial Peripheral Interface) and tested using USB SPI-Logic Analyzer Hantek4032L
(Qingdao Hantek Electronic, China). The acquired data of the 16-channel from this prototype has been successfully
transferred to a PC (Personal Computer) with accuracy greater than 91 %. The data acquisition system has been
visualized with time-domain format in the multi-graph plotter, the frequency-domain based on FFT (Fast Fourier
Transform) calculation, and also brain-mapping display of 16-channel. The GUI (Graphical User Interface) has been
developed based on OpenBCI (Brain Computer Interface) using Java Processing and also can be stored of data in the
*.txt format. Instrumentation systems have been tested in the frequency range of 1-50 Hz using MiniSim 330 EEG
Simulator (NETECH, USA). The validation process has been done with different frequency of 0.1 Hz, 2 Hz, 5 Hz, and 50
Hz, and difference voltage amplitudes of 10 μV, 30 μV, 50 μV, 100 μV, 500 μV, 1 mV, 2 mV and 2.5 mV. However, the
acquisition system was not optimal at a frequency of 0.1 Hz and for amplitude potentials of over 1 mV had differences of
the order 10 μV.

INTRODUCTION
An electroencephalogram (EEG) is medical instrument which are widely used in neuroscience research and
clinical practice for monitoring and describe the functionality of brain, whereas qEEG (Quantitative EEG) for
identifying specific purposing to visualize based on mathematical processing for their specific brain functions [1].
EEG instrument is an important biomedical diagnostic technique for medical including a portable EEG detection
system based on gain or filtering interleaved architecture [2]. EEG is much cheaper than fMRI (Functional Magnetic
Resonance Imaging), encourages small clinics and health-care facilities [3] Moreover, a portable EEG system can
record multichannel EEG signals with low-power consumption [4].
EEGs signals were recorded due to electrical activity of the brain including in telemedicine applications [4].
There are many applications based on EEG data acquisition, such as medical imagery technologies, the diagnostic
method of brain injury, the brain-computer interface (BCI) techniques and other signal processing methods [5]. Most
of the systems based on CPU (Central Processing Unit) compatible instruction set to processing element mechanism
[6], and portable EEG have to use ultra-low power on Integrated Chips (ICs) [7]. The ADS1299 (Texas Instrument,
USA) is an analog front-end chip which can be cascaded a daisy-chain configuration into the digital data format.
EEG acquisition system based on two daisy-chained has been successfully implemented using ADS1299, a TI-
MSP430 microcontroller 16-bit, and wireless communication and developed in a laboratory scale [8]. The custom

International Symposium on Current Progress in Mathematics and Sciences 2016 (ISCPMS 2016)
AIP Conf. Proc. 1862, 030149-1–030149-6; doi: 10.1063/1.4991253
Published by AIP Publishing. 978-0-7354-1536-2/$30.00

030149-1
platform the 8-channel data acquisition system based-on ADS1299 and MSP430 microcontroller have been verified
for BCI tasks [9] and the study of BCI has grown dramatically [10].
Basically, the ADS1299 can convert analog signals of EEG / ECG / EMG (biopotential measurement) into the
digital data format and equip with the filter element and signal gain [11]. In this study, the 16-channel EEG
acquisition system has been developed with daisy-chain technique 32-bit microcontroller ATSAM3X8EARM
Cortex-M3 using open source software of OpenBCI – Processing GUI (Graphical User Interface) application. On
previous, our research group result has been successfully developed a prototype system of data acquisition to 8-
channel EEG [12]. In this paper, the EEG prototype based on 16-channel electrodes would be the basis for further
study of the 32-channel data acquisition and also integrate with previous studies of stroke identification [13-15].

The Data Acquisition System

The block diagram of data acquisition of the developed EEG in this study is shown in Fig. 1. A daisy chain
technique is a technique to acquire more channels EEG signals of simultaneously that are implemented on the
ADS1299. This chip has an SPI (Serial Peripheral Interface) communication protocol to a digital processor. The
ADS1299 has eight channels with a simulaneous sampling of 24-bits, which is designed for biopotential applications
such as EEG, ECG, EAP (Audio Evoked Potential), BIS (Bispectral Index) and also Sleep Study. It has low-noise
differential input singles, PGA (Programmable Gain Amplifier) selectable to: 1, 2, 4, 6, 8, 12, and 24 with data rate
from 250 SPS to 16 KSPS. This chip has criteria low-power consumption of 5mW/channel, input bias current of 300
pA and -110dB CMRR, which works simultaneous and high-precision to acquire the EEG signal in multichannel
formats up to 8-channels per-chip elements [11,16]. That device was interchanged amplifier with differential
Instrumentation Amplifier (IA) and filter-stages mechanism with High Pass Filter (HPF) and also Dual-Notch Low
Pass Filter (DNLPF) [2], along with lead-off detection feature [17].

METHODS
The implementations of biomedical technology based on embedded processors with signal processing element
were in a wide range of applications [18]. In this study, the EEG acquisition-system was developed based on
embedded-system with FFT (Fast-Fourier Transform) signal processing capability. The embedded processor was
used based on ARM Cortex-M3 architecture, and suitable for mapping medical applications requiring ultralow-
power consumption using SPI protocol [19].
EEG signals can be considered as a realization of a random or stochastic process [20], therefore it needs
embedded-system with a processor based. The data acquisition techniquehas been widely applied based-on ARM
Cortex-M3 ATSAM3X8E and ADS1299, which provides powerful computations [21] and connected via a shared
bus to a discrete SPI controller [22].
The recorded biological signal conditioning needs a successive analog and digital transformation format, the
amplitude and the signal-to-noise ratio (SNR) of the recorded bio-signals require very low. The Common-Mode
Rejection Ratio (CMRR) of a differential amplifier as high as possible and a low value of the off-set of current drift
[23]. The FFT can be accelerated to extract the band powers mechanism such as the alpha, beta, and delta wave [13,
24, 25] and needs high computational requirements [26]. This study was focus on 16-channels EEG data acquisition
system, which was a development from previous studies [12, 15, 27].

FIGURE 1. The block diagram of EEG data acquisition based-on embedded-system

030149-2
 Standardization SPI Protocol
The standard mechanism to activate the ADS1299 based on the SPI communication protocol standard [11]. The
DOUT (Data Output) SPI generates the first MSB (Most Significant Bit) to the first LSB (Least Significant Bit)
format data. After 216 SCLK purpose collecting the data afterward the device-1 was reproduced the data acquisition
result, and also going to 217 SCLK the device-2 was reproduced the data from DAISY_IN0-DOUT1. The
mechanism method of daisy-chained SPI communication protocol can be seen in Fig. 2.

Mechanism of Multi-channel Data Acquisition

The daisy-chain configuration and ADS1299 architecture are shown in Fig. 3. The system consisted of 16-
electrodes placed on a human head. The two ADS1299 were connected via daisy-chain configuration through SPI
protocol to be connected directly to the ARM microcontroller ATSAM3X8E and the acquired data can be displayed
on a visual screen.

RESULTS AND DISCUSSION

The data-acquisitions system was developed into 16-channel and tested as shown Fig. 4. The research was
focused on the development of using two chips ADS1299 based on Daisy-chain configuration to obtain
simultaneous 16-channels. The Instrumentation system has successfully captured EEG signals of 1-60 Hz displayed
in real-time with their FFT counter parts. For each segment, the signal were averaged and displayed their power
spectrum[3]..

FIGURE 2. The SPI (Serial Peripheral Interface) standard protocol of ADS1299 Daisy-Chain configuration

(a) (b)
FIGURE 3. (a) The Daisy-Chain ADS1299 configuration and (b) the Architecture standard of ADS1299 front-end

030149-3
 The EEG data acquisition result may be confounded with different types of noises, such as eye blinks,
movements and muscular activates (e.g., the heart beats) [3], and also artifacts movement, half-cell effect, capacitive
and electrical signal of 50/60 Hz interference, and others [9]. To achieve a proper interference cancellation be able
from SNR (Signal Noise Ratio) [28]. The EEG data acquisition is suggested to use digitized mechanism at 250 SPS
(Sampling per Second), a high pass filter of 0.1 Hz, and also low pass filter of 30 Hz [29]. The perturbation effects
on the skull are more sensitive to measure the conductivity of the tissue of brain [30].
The SPI signal result ADS1299 front-end has been verified with USB SPI-Logic Analyzer Hantek4032L
(Qingdao Hantek Electronic, China) and this is similar with Texas Instrument SPI protocol. The data acquisition
system was implemented of OpenBCI (Brain Computer Interface) GUI (Graphical User Interface) Processing
Software based to display the data results. That GUI panel can display an EEG signal data acquisition in real-time
and simultaneous format. The data-acquisition system has been done and tested on the ADS1299 front-end module
consists of 16-channel and whole the process of acquisition can save and display to panel a GUI based.
In the specific average result of calibration for all channels (1-16 channels) are shown in Table 1 and Fig. 5.
Calibration error at frequency 2 Hz was < 7.27 % and at frequency 5 Hz was < 6.48 %. In the specific result as were
obtained the information of brain functionality of human with EEG Caps. The EEG-Data Acquisition system was

(b)

(a) (c)
FIGURE 4. (a) The data-acquisition system part elements based-on ARM Cortex-M3 ATSAM3X8E, (b) MiniSim 330 EEG
Simulator (NETECH, USA) and,(c) The electronics front-end ADS1299 which is made from our Research Group.

TABLE 1. The Data result of calibration with EEG Simulator

Frequency Input Error Deviation
Voltage Input Acquisition Conversion Deviation
EEG-Simulator Acquisition System
Simulator (µV) Result-GUI (Vrms) Vrms to µV (µV)
(Hz) (%)
10 3.48 9.83 0.20 2.01
30 20.11 28.44 1.56 5.21
2
50 33.16 46.89 3.11 6.22
100 65.58 92.73 7.27 7.27
10 3.77 10.65 0.65 6.48
30 21.63 30.58 0.58 1.93
5
50 35.71 50.49 0.49 0.98
100 70.97 100.35 0.41 0.41

030149-4
 FIGURE 5. The capture signal results of 5 Hz, 30 µV calibration process per channel (1-16 channels) from
MiniSim 330 EEG Simulator (NETECH, USA) (with ± 2 % accuracy)

successfully tested and the data results are presented in real-time with three parts, namely EEG waves, FFT, and
Brain Mapping. The data-acquisition results were interpreted of brain signals by ADS1299 based with µVrms and
FFT frequencies. The Raw Data of EEG signal are presented in Vrms unit and also can display the impedance values
for each electrode, that waveform was filtered with band-pass at 0.5-30 Hz with maximum deviation error 7.27 % at
2 Hz, and manually inspected to identify any movement or muscle artifact, electrical impedance fluctuation, cable
movements, bad contact of wire electrodes, and also conductivity influence from electrodes paste.

CONCLUSIONS
The EEG data acquisition prototype has been developed based on 32-bit microcontroller ARM Cortex-M3
ATSAM3X8E and Analog Front-end ADS1299 for acquiring a brain signal activity with accuracy greater than
91 %. An acquisition system was successfully implemented in a portable size and synchronous simultaneous
processing. The prototype has been implemented to configure an EEG data acquisition of 16-channels. The system
was implemented of using daisy-chain configuration protocol a SPI with the synchronous communication process.
The acquired signals, the FFT signal processing were displayed in a real-time graph and saved into data storage.

ACKNOWLEDGMENTS
This research was supported by Biomedical Research Grant DIKTI-Department of Education and Culture of the
Republic of Indonesia under contract number: 1720/UN2.R12/JKP.05.00/2016.

REFERENCES
1. J. M. Rogers et al., Int. J. Psychophysiol. 106, 87–96 (2016).
2. T. B. Nazzal, S. A. Mahmoud, and M. O. Shaker, Microelectronics J. 56, 81–96 (2016).
3. W. Mumtaz et al., Biomed. Signal Process. Control 31, 108–115 (2017).
4. L. Lin, Y. Meng, J. Chen, and Z. Li, Biomed. Signal Process. Control 20, 45–51 (2015).
5. B. Albert et al., Procedia Comput. Sci. 96, 703–712 (2016).
6. K. H. Lee, S. Member, and N. Verma, IEEE J. Solid-State Circuits 48, 1625–1637 (2013).

030149-5
7. P. J. Davies and J. Bohórquez, Proceedings of the 29th Southern Biomedical Engineering Conference, Miami,
2013 (IEEE, New Jersey, 2013), pp. 63–64.
8. Y. Zou and V. Nathan, IEEE J. Biomed. Heal. Inform. 20, 73–81 (2016).
9. V. Nathan and R. Jafari, IEEE Trans. Biomed. Circuits Syst. 9, 631–640 (2015).
10. J. Minguillon, M. A. Lopez-gordo, and F. Pelayo, Biomed. Signal Process. Control 31, 407–418 (2017).
11. Texas-Instrument, “Low-Noise , 8-Channel , 24-Bit Analog Front-End for Biopotential Measurements
ADS1299,” (2012), available at http://www.ti.com/product/ADS1299
12. L. O. H. Z. Toresano et al., Proceedings of the First International Symposium of Biomedical Engineering
(ISBE), Depok, 2016, edited by Y. Whulanza et al. (AIP Publishing, New York, 2017), Vol. 1817, p. 040009.
13. O. N. Rahma, S. K. Wijaya, Prawito, and C. Badri, Proceedings of the 9th IASTEM (International Academy of
Science, Technology, Engineering and Management) International Conference, Bali, 2016 (World Research
Library, 2016), pp. 17–22.
14. E. P. Giri, A. M. Arymurthy, and S. K. Wijaya, Proceedings of the 2015 International Conference on
Advanced Computer Science and Information Systems (ICACSIS), Depok, 2015 (IEEE, New Jersey, 2015), pp.
297–301.
15. M. I. Sudrajat, S. K. Wijaya, H. Arjadi, and S. Abdan, J. Tek. Biomed. Ind. 2, 1-7 (2016).
16. D. Acharya, A. Rani, and S. Agarwal, Proceedings of the 4th International Conference on Reliability, Infocom
Technologies and Optimization (ICRITO), Noida, 2015 (IEEE, New Jersey, 2015).
17. E. Mastinu, M. Ortiz-catalan, and B. Håkansson, Proceedings of the 37th Annual International Conference of
the IEEE Engineering in Medicine and Biology Society (EMBC), Milan, 2015 (IEEE, New Jersey, 2015).
18. Z. Song, Z. Ji, J. Ma, B. Sputh, U. R. Acharya, and O. Faust, Comput. Methods Programs Biomed. 108, 656-
664 (2012).
19. S. R. Sridhara et al., IEEE J. Solid-State Circuits 46, 721–730 (2011).
20. V. Gandhi et al., IEEE Trans. Neural Networks Learn. Syst. 25, 278–288 ( 2014).
21. C. Lin, B. Lin, F. Lin, and C. Chang, IEEE Syst. J. 8, 363–370 (2014).
22. A. Page et al., IEEE Trans. Circuits Syst. - II Express Briefs 62, 109–113 (2015).
23. L. Losonczi, L. F. Márton, T. S. Brassai, and L. Farkas, Procedia Technol. 12, 141–147 (2014).
24. T. Roh et al., IEEE Trans. Biomed. Circuits Syst. 8, 755–764 (2015).
25. Y. Dai, X. Wang, X. Li, and Y. Tan, Measurement 74, 11–20 (2015).
26. H. Woehrle, J. Teiwes, E. Kirchner, and F. Kirchner, APCBEE Procedia 7, 60–66 (2013).
27. A. Syakura, Designing eight channels Electroencephalogram (EEG) system based on ADS1299 controlled by
Arduino uno for signal decomposition, BSc Project, Universitas Indonesia, Depok, 2015.
28. M. Tomasini et al., IEEE Sens. J. 16, 3887–3895 (2016).
29. I. Käthner et al., Clin. Neurophysiol. 124, 327–338 (2013).
30. J. Li, K. Wang, and S. Zhu, Phys. Med. Biol. 49, 701–717 (2004).

030149-6