JNEEG

NEEG: Turn Your Jetson Nano into a Brain-Computer Interface with a GPU on Board

Measure brain (EEG), muscle (EMG) and heart (ECG) signals with research-grade 24-bit precision, and run neural networks on them in real time, right on the NVIDIA Jetson Nano. JNEEG is an open-source 8-channel shield that plugs straight onto the Jetson Nano 40-pin header. Record, process and classify brain signals on one device, with no cloud and no extra computer.

Note: the Jetson Nano, electrodes, EEG cap and power bank are not included. See "What you also need" below.

Why JNEEG?

• Machine learning on the device. The Jetson Nano's GPU runs neural networks on live EEG: real-time classification, neural decoding, or EEG combined with a camera.
• Research-grade signal quality. Built on the Texas Instruments ADS1299, the same 24-bit analog front end used in professional EEG systems.
• Plug and play. Mounts directly on the Jetson Nano 40-pin header. No wires, no soldering.
• Easy programming. Ready-made Python scripts for live graphs and saving data, plus support in PiEEG Server.
• Fully open source. Documentation, Python code and an example dataset on GitHub. No subscriptions, no locked software.
• Flexible. Use dry or wet electrodes, record EEG, EMG or ECG, and change sample rate and gain to suit your experiment.

What you can do with it

• Train and run neural networks on EEG with PyTorch or TensorRT, in real time
• Build an eyes-closed detector from alpha brain waves
• Use blinks and jaw clenches as buttons to control robots, games or devices
• Combine EEG with a camera in one AI application
• Record muscle (EMG) and heart (ECG) signals
• Teach neuroscience, signal processing and AI with real hardware

Features

• 8 biosignal channels, plus reference and bias electrode inputs
• TI ADS1299 analog front end, 24-bit resolution, all channels sampled simultaneously
• Sample rate from 250 SPS to 16 kSPS
• Programmable gain: 1, 2, 4, 6, 8, 12, 24
• Impedance measurement (lead-off detection) to check electrode contact
• Works with dry and wet (Ag/AgCl) electrodes via a standard 2.54 mm header
• Plugs directly onto the Jetson Nano 40-pin GPIO header (SPI interface)
• Low-noise power regulators on board
• LEDs for power and ADS1299 status
• Easy programming in Python

Specifications

Package includes

• 1 x JNEEG shield

What you also need

• NVIDIA Jetson Nano Developer Kit with a microSD card and JetPack
• 5 V power bank, 3 A output, up to 10,000 mAh
• 8 EEG electrodes (dry or wet) with 2.54 mm connectors, plus 2 ear-clip electrodes for reference and bias. The 8-channel EEG cap kit includes all of them.
• Battery-powered monitor, wired keyboard and mouse

Get started in 4 steps

1. Flash JetPack to the microSD card and enable SPI with the Jetson-IO tool.
2. Plug JNEEG onto the Jetson Nano (switched off) and connect the electrodes and ear clips.
3. Power the Jetson Nano from a power bank. The LED on the shield lights up.
4. Run the live-graph Python script and watch your brain signals.

Step-by-step guide: docs.pieeg.com/hardware/devices/jneeg

Real signals

20 seconds of 8-channel EEG recorded with JNEEG, band-pass 1-40 Hz. The large bursts are artifacts such as blinks or chewing.

Documentation & Links

• JNEEG Documentation: setup, wiring, software, specifications, troubleshooting
• JNEEG on GitHub: Python scripts and example dataset
• PiEEG Server: real-time streaming, dashboard and recording
• Paper on arXiv: Device JNEEG to convert Jetson Nano to brain-computer interfaces
• pieeg.com · Community on Discord

Important safety notice

JNEEG is not a medical device and is intended for research, education and engineering only. It must be powered from a 5 V battery (power bank) only. The Jetson Nano and every device connected to it (monitor, keyboard) must be battery-powered while electrodes are on a person. Never use the Jetson's wall power supply, mains power or an Ethernet cable. Please read the liability notice and the safety guide before use.