
FreeEEG32-alpha131
neuroidss/FreeEEG32-alpha131
https://hackaday.io/project/181683-freeeeg32-alpha131

Open-hardware 128-channel EEG acquisition system — design audit, firmware rewrite roadmap, Pi Zero 2W host stack, and synthesis paper
FreeEEG128-alpha is an open-source printed circuit board by m9h, from the m9h/freeeeg128 repository on GitHub. It is a 4-layer board of 70 × 70 mm with 168 parts on 505 nets, routed with 6,197 tracks and 359 vias.
It is built around the ADS131M08IRSNR, SN6505BDBV and LP5907, with the ADUM1402BRWZ, ADUM3160BRW and ISO7341CDW among its other parts. Its parts include 96 capacitors, 28 ICs, 21 connectors, 10 resistors and 2 diodes. It belongs with the STM32 boards of this library.
From Nunez and Gevins to Tucker's Geodesic Sensor Net to MindBigData — a lineage of high-density EEG, and a modern rewrite for 2026.
A design audit, firmware-rewrite roadmap, host-side Python stack, and synthesis paper for an open-hardware 128-channel scalp electroencephalography system. The hardware is the NeuroIDSS FreeEEG128-alpha: a STM32H743ZI microcontroller driving sixteen Texas Instruments ADS131M08 24-bit Δ-Σ ADCs across six SPI buses, with a medical-grade digital-and-power-isolation barrier between subject and host. Fewer than five alpha units exist in the world; one is on the bench where this repository is being written.
The goal: rewrite the firmware, ship the host-side software stack, validate against Black et al. 2017 (Open Ephys + EEG), and plan the beta revision — STM32H743 acquisition + Raspberry Pi Zero 2W companion in a shoulder-bag form factor — with first-class support for LSL, BrainFlow, Open Ephys GUI, MNE-CPP, and EEG-ExPy on day one.
timeline
title Four decades of high-density EEG
1981 : Nunez — Electric Fields of the Brain
1980–86 : Gevins — EEG Systems Laboratory (1980) & SAM Technology (1986)
1993 : Tucker — Geodesic Sensor Net
1994 : Nunez & Westdorp — Surface Laplacian & high-resolution EEG
1996 : Pflieger & Sands @ Neuroscan — FIRST 256-channel EEG
Mid-1990s: EGI builds the SECOND 256-ch system (Morgan Hough, first SW-eng manager)
Late 1990s: BioSemi ActiveTwo — first HD-EEG with active electrodes
2006 : Nunez & Srinivasan — 2nd ed. of Electric Fields of the Brain
2014 : OpenBCI Cyton — open-source EEG hits the mainstream
2017 : Open Ephys + EEG (Siegle et al., Black et al.) — 128-ch open platform
2021 : NeuroIDSS FreeEEG32-beta — ADS131M08 proving ground
2022 : Fiedler et al. — 256-channel dry cap for VHD-EEG
2023 : NeuroIDSS FreeEEG128-alpha — 16× ADS131M08, MindBigData 2023 MNIST-8B
2024 : g.tec g.Pangolin (Schreiner et al.) — 1024-channel VHD-EEG SOTA, matches ECoG
2024 : Neurogate OctaFlow — ESP32-C3 + ADS1299, 8-ch teaching board
2026 : FreeEEG128 firmware rewrite + Pi-Zero-2W beta (this project)The full story — Nunez's biophysical arguments for why >100 electrodes matter, Gevins' real-time cognitive-state engineering at SAM Technology, Tucker's five-minute saline-sponge application trick at EGI, Ramesh Srinivasan as the conduit that carried Nunez's framework into EGI's product line, the commercial and open-hardware generations — lives in HISTORY.md.
flowchart LR
subgraph CAP["cap on head"]
ELEC["130 dry Ag/AgCl<br/>passive, 10/5 system<br/>1.5mm DIN plugs"]
BB["brain-box<br/>passive fan-out"]
end
ELEC --> BB
subgraph BAG["shoulder-bag amp"]
PCB["FreeEEG128 PCB<br/>STM32H743ZI + 16× ADS131M08<br/>ADuM medical isolation"]
PI["Raspberry Pi Zero 2W<br/>Pi OS 64-bit Lite<br/>pylsl · MNE · BrainFlow · EEG-ExPy"]
BAT["2S 18650 Li-ion pack"]
PCB <--> |"USB-CDC"| PI
BAT -.-> PCB
BAT -.-> PI
end
BB --> |"HD D-Sub HD78<br/>30–60 cm shielded bundle"| PCB
PI --> |"Wi-Fi<br/>LSL · WebSocket · FastAPI UI"| HOST["laptop<br/>LabRecorder · MNE Scan<br/>OpenViBE · Timeflux"]
LS["Neurobs Labstreamer<br/>TTL + photodiode"] --> PIThe STM32H743 is retained across alpha → beta because no other MCU in this class has six general-purpose SPI peripherals plus the TIM1/TIM8-cascaded-DMA-to-GPIO primitive that generates sample-synchronous chip-select pulses across sixteen ADCs with zero CPU involvement. The Pi Zero 2W companion replaces a planned ESP32-S3 co-processor; it eliminates a 2-to-4-week second-firmware project, gives us the full Python stack native on-device, and reduces the STM32↔host link to plain USB-CDC. Details and trade-offs in COMPARISON.md and paper/paper.pdf.
We treat Black, Voigts, Agrawal, Ladow, Santoyo, Moore & Jones (2017), "Open Ephys electroencephalography (Open Ephys + EEG): a modular, low-cost, open-source solution to human neural recording", J. Neural Eng. 14:035002 (doi:10.1088/1741-2552/aa651f) as the primary feature-for-feature comparison benchmark. Their system is built on the Open Ephys acquisition board (Opal Kelly XEM-6310 FPGA + Intan RHD2132/RHD2164 16-bit headstages) and validated against a BrainVision actiCHamp clinical amplifier using eyes-closed 8–14 Hz alpha — the paradigm we will replicate as our beta acceptance gate.
| Open Ephys + EEG (Black 2017) | FreeEEG128 | |
|---|---|---|
| ADC | Intan RHD2132/2164, 16-bit | 16× TI ADS131M08, 24-bit |
| Channels | up to 128 (2× RHD2164) | 128 fixed |
| Isolation | partial (inherits Open Ephys board) | full medical-grade on-board |
| Form factor | desk/rack | belt-pack wearable |
| Cost (system) | ~USD 2000–3500 | ~USD 500–1000 BOM |
| License | CC-BY-NC-SA 3.0 IGO (non-commercial) | AGPL-3.0 + CERN-OHL-S-v2 (planned) |
| Ecosystem fit | Open Ephys GUI native | LSL / BrainFlow / MNE-CPP / Open Ephys GUI |
Full comparison matrix in COMPARISON.md.
The very-high-density (VHD) EEG state-of-the-art is now g.tec's g.Pangolin, a 1024-channel-capable platform built on flex-PCB active-wet electrode grids with 8.6 mm spacing and on-grid pre-amplification, feeding up to four 256-channel g.HIamp amplifiers. Schreiner et al. (2024, Scientific Reports, doi:10.1038/s41598-024-57167-y) demonstrate central-sulcus delineation at 95.2% accuracy — comparable to invasive intracranial ECoG.
FreeEEG128 is deliberately not chasing VHD status. Open hardware has no peer at 256+ channels as of 2026; we occupy the "affordable, open, 128-channel wearable" niche — an order or two of magnitude below g.Pangolin in both channels and cost, above everything else in the open-hardware column. A natural FreeEEG256 upgrade on the same STM32H743 + ADS131M08 chassis is plausible (32 chips across 6 SPI buses) but out of scope for this project. See HISTORY.md §4b and COMPARISON.md for the VHD-EEG SOTA treatment.
| Path | What it is |
|---|---|
paper/ | Synthesis paper — Sweave (paper.Rnw) + refs.bib + Makefile → tectonic → paper.pdf |
host/ | Python host stack for the Pi Zero 2W: benchmarks, LSL outlet, capture client (scaffolded) |
HARDWARE.md | Board, cap, connectors, host link, setup/QC procedure |
PERIPHERALS.md | STM32 peripheral inventory + firmware-rewrite service list |
COMPARISON.md | Open-hardware EEG landscape + Black 2017 feature matrix |
ENCLOSURE.md | Belt-pack form factor, battery, connector, cable management |
HISTORY.md | Nunez → Gevins → Tucker/EGI → open-hardware lineage |
ROADMAP.md | Alpha firmware rewrite + beta design; MCU decision rationale |
REFERENCES.md | Curated EEG-hardware-build bibliography |
Upstream repositories (AGPL-3.0, referenced rather than redistributed):
*.md files and the synthesis paper.ROADMAP.md. Wire-packet format is the first artefact to lock because it is the alpha → beta carry-over invariant.host/ and can be run the moment a Pi arrives. BrainFlow's SYNTHETIC_BOARD stands in for the real device until the firmware ships; when it does, only the transport backend changes.Validation gate before anything ships as "beta": reproduce Black et al. 2017 — eyes-closed 8–14 Hz alpha, side-by-side vs a BrainVision actiCHamp, same average-power and SNR metric.
Read the paper first. paper/paper.pdf is the single-document synthesis of everything in this repository (~30 pages).
Pi Zero 2W parallel track (while the alpha firmware rewrite proceeds):
# on the Pi, after flashing Raspberry Pi OS 64-bit Lite
sudo apt install -y git
git clone https://github.com/m9h/freeeeg128.git
cd freeeeg128/host
./scripts/install_pi.sh
source ~/freeeeg/bin/activate
make benchBenchmarks land in host/results/ as timestamped CSVs. See host/README.md.
Rebuild the paper locally (requires R, tectonic):
cd paper && makeThe intellectual and engineering lineage of this project is long. Paul Nunez and Alan Gevins laid the theoretical and practical foundations of high-density and high-resolution EEG. Don Tucker built the Geodesic Sensor Net and with it the form factor this system inherits; Ramesh Srinivasan carried the Nunez biophysical framework into EGI's product line. M.E. Pflieger and S.F. Sands at Neuroscan built the first 256-channel system. The entire open-hardware generation — Joel Murphy and Conor Russomanno at OpenBCI, Josh Siegle and Jakob Voigts at Open Ephys, Christopher Black and colleagues on the Open Ephys + EEG paper, Adam Feuer at HackEEG, Gaetano Gargiulo at BIOADC, Ildar Rakhmatulin at PiEEG/IronBCI, David Vivancos at NeuroIDSS/FreeEEG — made this project thinkable.
AGPL-3.0 for code, firmware, and documentation in this repository. The hardware portion is currently AGPL-3.0 (inherited from the upstream NeuroIDSS FreeEEG128-alpha); migration to CERN-OHL-S-v2 is planned for the beta revision. See LICENSE.
Third-party licenses are respected: Intan, ADI, TI, and ST datasheet material cited under fair use; the paper and docs cite upstream authors directly.
Built in collaboration with Claude (Anthropic).
Because the most interesting EEG system to build is the one you understand top to bottom.
| Qty | Value | Footprint | References |
|---|---|---|---|
| 36 | 1u | Capacitor_SMD:C_0402_1005Metric | C4, C10, C12, C13, C15, C22, C24, C27, C29, C32, C34, C46, C48, C100, C101, C103 +20 |
| 29 | 100n | Capacitor_SMD:C_0402_1005Metric | C1, C2, C3, C20, C21, C25, C35, C36, C37, C38, C39, C40, C41, C42, C43, C44 +13 |
| 16 | 220nF | Capacitor_SMD:C_0402_1005Metric | C5, C8, C11, C14, C23, C28, C33, C78, C81, C84, C87, C90, C93, C96, C99, C102 |
| 16 | ADS131M08IRSN | Package_DFN_QFN:QFN-32-1EP_4x4mm_P0.4mm_EP2.65x2.65mm_ThermalVias | U101, U102, U103, U104, U105, U106, U107, U108, U109, U110, U111, U112, U113, U114, U115, U116 |
| 8 | 2.2u | Capacitor_SMD:C_0603_1608Metric | C16, C17, C18, C19, C53, C58, C75, C76 |
| 8 | Conn_02x20_Odd_Even | Connector_PinHeader_1.27mm:PinHeader_2x20_P1.27mm_Vertical_SMD | J201, J202, J203, J204, J205, J206, J207, J208 |
| 4 | 24 | Resistor_SMD:R_0402_1005Metric | R113, R114, R115, R116 |
| 4 | Conn_01x02 | Connector_PinHeader_2.54mm:PinHeader_2x01_P2.54mm_Horizontal | J100, J101, J104, J105 |
| 3 | 0 | Resistor_SMD:R_0402_1005Metric | R109, R110, R112 |
| 3 | LP5907 | Package_TO_SOT_SMD:SOT-23-5 | U6, U8, U22 |
| 2 | 10u | Capacitor_SMD:C_0603_1608Metric | C68, C70 |
| 2 | 4k7 | Resistor_SMD:R_0402_1005Metric | R84, R85 |
| 2 | CL1 | Capacitor_SMD:C_0402_1005Metric | C64, C66 |
| 2 | CL2 | Capacitor_SMD:C_0402_1005Metric | C65, C67 |
| 2 | Conn_02x02_Odd_Even | Connector_PinHeader_2.54mm:PinHeader_2x02_P2.54mm_Horizontal | J1, J2 |
| 2 | Conn_02x03_Odd_Even | Connector_PinHeader_2.54mm:PinHeader_2x03_P2.54mm_Horizontal | J81, J84 |
| 2 | Conn_02x04_Odd_Even | Connector_PinSocket_1.27mm:PinSocket_2x04_P1.27mm_Vertical | J301, J303 |
| 2 | D_Schottky | Diode_SMD:D_SOD-123 | D1, D2 |
| 2 | SW_Push | Button_Switch_SMD:SW_SPST_B3U-1000P-B | SW1, SW2 |
| 2 | USB_B_Micro | Connector_USB:USB_Micro-B_Molex-105017-0001 | J99, J103 |
| 1 | 16MHz | smd5032:Crystal_SMD_EuroQuartz_MJ-4Pin_5.0x3.2mm_HandSoldering | Y2 |
| 1 | 32768Hz | Crystal:Crystal_SMD_EuroQuartz_EQ161-2Pin_3.2x1.5mm_HandSoldering | Y1 |
| 1 | 4.7u | Capacitor_SMD:C_0603_1608Metric | C51 |
| 1 | 473521001 | 473521001:473521001 | U21 |
| 1 | 750315371 | 750315371:750315371 | T1 |
| 1 | ADUM1402BRWZ | Package_SO:SOIC-16W_7.5x10.3mm_P1.27mm | U16 |
| 1 | ADuM3160 | Package_SO:SOIC-16W_7.5x10.3mm_P1.27mm | U17 |
| 1 | Conn_02x04_Odd_Even | Connector_PinHeader_1.27mm:PinHeader_2x04_P1.27mm_Vertical | J302 |
| 1 | FXO-SM7_OSC-SMD7050_short | Oscillator:Oscillator_SMD_EuroQuartz_XO91-4Pin_7.0x5.0mm_HandSoldering | U7 |
| 1 | ISO7341c | Package_SO:SOIC-16W_7.5x10.3mm_P1.27mm | U15 |
| 1 | LSM6DS3 | LSM6DS3TR:Bosch_LGA-14_3x2.5mm_P0.5mm | U1 |
| 1 | REF2025 | Package_TO_SOT_SMD:SOT-23-5_HandSoldering | U2 |
| 1 | R_EXT | Resistor_SMD:R_0402_1005Metric | R117 |
| 1 | SN6505B | Package_TO_SOT_SMD:SOT-23-6_Handsoldering | U19 |
| 1 | STM32H743ZITx | Package_QFP:LQFP-144_20x20mm_P0.5mm | U18 |
The 162 most common of 168 parts.

neuroidss/FreeEEG32-alpha131
https://hackaday.io/project/181683-freeeeg32-alpha131

neuroidss/FreeEEG32-alpha1.5
4xAD7771BCPZ 64 kSPS, ADR4525BRZ 2.5V REF +-0.02%, STM32H743ZIT6 MCU 1027 DMIPS


neuroidss/FreeEEG32-beta
as FreeEEG32 plus alpha1.5 test passed, FreeEEG32 project changed to beta version

neuroidss/FreeEEG32-alpha1.0
ALPHA1.0 TEST FAILED, USE NEW VERSION INSTEAD FreeEEG32 alpha1.0 32 channels mobile electroencephalography under GNU AGPL 24 bits, 4kSPS ADCs 4x AD7770 MCU STM32F427ZI WIFI ESP-12F (ESP8266) USB CP2102N Accel & hyro…



It measures 70 × 70 mm and has 4 copper layers, 4.48 mm thick.
168 parts, built around the ADS131M08IRSNR, SN6505BDBV, LP5907 and ADUM1402BRWZ. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.
From the m9h/freeeeg128 repository on GitHub, at the commit this page was made from: the board is pcb/kicad/FreeEEG128-alpha.kicad_pcb and its schematic pcb/kicad/ACCEL&GYRO.kicad_sch. PCBKit opens the board in the browser, with nothing to install.
Its repository states no licence, so its author keeps all rights: ask m9h before reusing the design.