Skip to main content

​
System Architecture

NeuroFocus has two main components: hardware and firmware. This page explains how they work together.

​
Hardware Architecture

​
Signal Chain

Electrodes → AD8422 → ADS1220 → ESP32-S3 → BLE/USB
1. Electrodes
  • Capture electrical brain activity
  • Typical signal: 1-100 µV
2. AD8422 Instrumentation Amplifier
  • Gain: 100x (configured with 100Ω resistor)
  • Amplifies weak EEG signals to measurable levels
  • High common-mode rejection removes noise
3. ADS1220 ADC
  • 24-bit resolution
  • Sampling rate: 660 SPS (default)
  • External 3.3V reference
  • Differential input mode
4. ESP32-S3 Microcontroller
  • Seeed Studio Xiao ESP32-S3
  • Handles SPI communication with ADC
  • Processes and streams data
  • Supports BLE and USB Serial

​
Power Supply

  • Input: USB 5V or battery
  • Regulator: TPAP2112K-3.3 LDO
  • Output: 3.3V for all components

​
Firmware Architecture

The V4 firmware uses a component-based design. Each component handles one responsibility.

​
Components

ADS1220Driver
  • Low-level SPI communication
  • ADC initialization and configuration
  • Data ready detection
  • Raw data reading
BLEManager (optional)
  • Bluetooth setup and management
  • Connection handling
  • Data transmission via notifications
  • Command reception
EEGStreamer
  • High-level streaming coordination
  • Routes data to Serial or BLE
  • Manages streaming state
CommandHandler
  • Processes commands from Serial and BLE
  • Unified interface for both sources
  • Returns command enums
LedController
  • Status LED management
  • Visual feedback for system state

​
Data Flow

ADS1220 → ADS1220Driver → EEGStreamer → BLE/Serial
                                ↓
                          CommandHandler
  1. ADS1220Driver checks for new data
  2. EEGStreamer reads raw ADC values
  3. Data streams to Serial and/or BLE
  4. CommandHandler processes user commands
  5. Commands control streaming state

​
Communication Modes

​
BLE Mode

When BLE is enabled:
  • Auto-starts streaming on connection
  • Auto-stops on disconnection
  • Commands from Serial or BLE
  • Data to both Serial and BLE

​
Serial Mode

When BLE is disabled:
  • Manual command control only
  • Data to Serial only
  • Smaller firmware size
  • Lower power consumption

​
Design Patterns

Singleton Pattern
  • BLEManager uses static instance for callbacks
Facade Pattern
  • EEGStreamer simplifies complex operations
Strategy Pattern
  • CommandHandler works with multiple sources
Dependency Injection
  • Components accept dependencies via constructor

​
Signal Calculations

​
Electrode to ADC

Input voltage at electrode: V_electrode After AD8422 amplification: 
V_amplified = V_electrode × 100
Example: 50 µV signal becomes 5 mV

​
ADC Output

The ADS1220 converts voltage to a 24-bit signed integer: 
ADC_value = (V_amplified / V_ref) × (2^23 - 1)
Full-scale range: ±3.3V

​
Expected Values

For typical EEG signals:
Signal (µV)Amplified (mV)ADC Value
101.0~2,533,000
505.0~12,665,000
10010.0~25,330,000

​
Memory Management

All components use stack allocation. No dynamic memory means predictable usage and no fragmentation. The BLE stack is managed by the ESP32 library with automatic buffer management.

​
Thread Safety

Key volatile variables:
  • deviceConnected - Modified in ISR
  • commandAvailable - Modified in callback
  • lastCommand - Modified in callback
The firmware uses flags to detect connection state changes safely.

​
Error Handling

Initialization Failures
  • Critical failures halt with LED blink pattern
  • Non-critical failures log to Serial
Runtime Errors
  • Graceful degradation
  • BLE send fails silently if disconnected
  • State validation before operations