Portable BCI Stimulator Final Presentation Group: 17 Bonnie Chen, Siyuan Wu, Randy Lefkowitz TA: Ryan May ECE 445 Monday, April 29 th, 2013
Overview Introduction Features BCI/EEG System Overview Design of Individual Modules Testing and Verifications Future Development Sponsors
Introduction Most brain computer interfaces (BCI) limited to laboratory settings We would like to help make the EEG system more portable through bluetooth Allows people to communicate without any type of movement
Features Portability –Wearable Size –Rechargeable Battery Wireless control –Bluetooth –Compatible with most computers Variable frequency and Intensity –Set by User
BCI/EEG System Overview
Stimulator Top Level
Computer/Wireless Module Overview Wireless communication between PC and Bluetooth Module through terminal Retrieve LED number, frequency and intensity level from user Check the validity of command
Wireless Transmitter/Receiver Built in Bluetooth 2.0 communication Standard TTL Bluetooth receiver Data sent wirelessly from PC to Arduino
Receiver – Arduino Diagram
Input Flowchart
Microcontroller Module Calculates Runtime Determines if each LED should toggle Sets LED values Latches values into LED driver
Timing Flowchart
Dividing Interval for On and Off Calculate LED state time (on/off) –Interval = current – previous Compare to Required Time –1/(2*frequency) Toggle if needed Save new time –Previous = current
TL5940 LED Driver Overview 16 Output Channels Rref = 2k ohms Intensity set by PWM Frequency controlled by Arduino TLC5940 library
TLC-5940 Library TLC.set(channel,intensity); –Loads TLC Register Tlc.update(); –Latches data into LED driver
TLC-5940 Arduino Connections
7.4V Power Supply Venom 1250mAh 10C 7.4V Lithium Ion Battery
LED Array Powered by 5V output from Arduino Flashes at frequency values between 1-9 Hz based on Arduino Code LED Intensities based on PWM values from LED Driver 5-10 LEDs mounted on adjustable frame
Lilypad Micro LEDs 3.3mm long Forward Voltage of V 200mA forward current Power Dissipation of 120mW
PCB Schematic
PCB Design Top Bottom
Final Design
Demo with the EEG
Review of Requirements Wireless Communication Portability Sufficient Power Successful Classification over different frequencies on EEG System
Testing and Verifications EEG Classification Frequency Bandwidth Power Budget
EEG Classification Demo Frequencies –6, 7, 8, 9 Hz Classification –All 4 Frequencies classified correctly (within 0.3Hz) Intensity of 20 out of 4096 –Fast Response –Comfortable Viewing
Frequency Bandwidth 1.Record LED Driver Output on Oscilloscope 2.Analyze EEG data with MATLAB 3.Compare Variance with EEG Classifier Sensitivity 4.Adjust Sensitivity values of EEG Program Accordingly 5.Test user response on EEG with updated sensitivity values
Frequency Bandwidth Cont. 1 Hz: μ = , σ 2 = Hz: μ = , σ 2 = Hz: μ = , σ 2 = Hz: μ = , σ 2 = Hz: μ = , σ 2 = Hz: μ = , σ 2 = Higher Frequencies produced less stable results SSVEP measurements generally 5-15 Hz Less accurate frequencies cause slower EEG response times
Frequency Bandwidth Cont.
Power Budget ComponentImax (mA)Voltage Microcontroller40 * 2 output pins7-12V (ideal) LEDs20 * (10 LEDs)3.2V (green, white) Bluetooth Module403.3V LED Driver1205V __________________________________________________ Total Estimated Usage time = 1250 [mAh] / 440 [mA] ≅ 3 hours of charge Factors to consider: - PWM value will never be over 50% (the blinking LED is on less than half of the time) - Able to get same results using 5 LEDs instead of 10 LEDs
Future Development Safer operating limits for near-eye LEDs Determine ideal threshold for response time, classification, and stability of the system. Improvements on mounting frame mechanics (aesthetics and functionality) Use Feedback from the EEG to implement commands that can control a range of devices (Quadcopter, Paralysis Assistance)
Sponsors A special thanks to the following people who helped make this project happen James Norton (Beckman) Erik Johnson (Beckman) David Jun (Beckman) Ryan May Professor Carney The friendly folks in the ECE parts shop
Thanks!!