Which architecture to choose if we want to build our own 32 channel EEG device?

Last Updated on September 22, 2024 by pg@petergamma.org

Indonesian scientists around La Ode Husein Z. Toresano have written several papers how to build a 32 channel EEG device with the ADS1299 TI demo kit:

  • they used four ADS1299 TI demo kits and a single board computer for 32 EEG channels.
  • they used daisy chain algorithm to read out the EEG data.
  • This code has not been published by the Indonesian Scientists
  • There are issues with this solution that it is not save for humans:

https://www.eevblog.com/forum/testgear/how-to-solve-the-ti-ads1299-eeg-demo-kit-is-not-save-for-humans-problem

Adam Feuer’s HackEEG 32:

  • it is based on the Arduino Due and four ADS1299 EEG shields
  • We have seen only a demo with 8 channels.
  • But the hardware is there, and also software examples.
  • But we do not know of any soft- and hardware tests with 32 EEG channels
  • The HackEEG is not available anymore, but it is open source
  • There is a 32 channel version of OpenBCI GUI which can be used with HackEEG 32

The Beats device with 32 channels developed by Chinese developers

  • It is similar to HackEEG 32
  • It uses a Raspberry pi 4 and four ADS1299 EEG Raspberry Pi shields
  • There are two papers available
  • It uses daisy chain code which is open source
  • The whole device is open source.
  • We do not know of independent tests of this device
  • We do not know if the Raspberry Pi 4 is the best SBC to build such a device.
  • Eventually it could be tested with a Raspberry Pi zero if there is issues with noise.
  • It is not on the market, but all components are open source and could be tested further

PiEEG

  • The PiEEG shield is new and available.
  • If we want to build a 32 channel device, this architecture would be suitable in principle, as we described previously.
  • PiEEG is especially suitable to do some tests what can be done with it ,and also to find the best single board computer,
  • Or if nothing works we can also try to solder a microcontroller to it.
  • We first have to have an architecture which works without issues, and is free from EEG noise.
  • We only know of little tests about PiEEG
  • There is Pyton code available
  • PiEEG is especially suitable to do some experiments.

To modify the OpenBCI architecture which is free from noise and well tested

Own printed circuit board with our own single board computer or micro controller

  • this is also an option
  • PiEEG developer Ildar Rakmulding has shown this on his YouTube channel how to let develop our own printet circuit board with PCBway

None of these 32 channel variants are on the market. 32 channel devices for instance from mBrainTrain are on the market and expensive. We suppose that most variants discussed here are abandoned prototypes which have issues. So we have to go deeper and do our own experiments until we got a solution which fit’s for our needs. And also we have to study if the ADS1299 is the right solution for us personally, or if we should choose a solution based on InfluxDB.