After studying the architecture of the ADS1299 EEG chip from Texas Instruments we came to the conlusion that this chip is suitable for building EEG devices for medical and research grade applications up to a large number of channels.
OpenBCI
- uses ADS1299
- is limited to 16 channels
- limiting is the OpenBCI architecture
- We could use in principle 2 x 16 OpenBCI modules to have 32 channels.
- But these modules cannot be connected to each other directly, what would be necessary for this purpose.
- For high quality standards all channels need to be sampled at exactly the same time.
- This is not possible with the OpenBCI architecture, since there is no direct connection between two 16 channel EEG modules.
- OpenBCI is on the market since 10 years.
- It has proofen to be free from EEG noise.
- It uses a PIC32MX250F128B microcontroller which uses very little power
- We suppose that the PIC32MX250F128B causes only little or no noise in the EEG signal.
HackEEG
- HackEEG was sold as field-proven, and successfully used for neuroscience research in several organizations, among those several universities:
https://www.crowdsupply.com/starcat/hackeeg
- But it was only temporarily available until it was pulled from the market again.
- It offered a 32 channel version.
- This was advertised also on “Mouser Switzerland”, as Entwicklungstools für multifunktionale Sensoren HackEEG 32-channel Deluxe Kit
- Here the hardware configuration for HackEEG 32:
https://docs.starcat.io/hackeeg-crowdsupply/user/hardware-32-channel.html
- Here the “Software Installation – 32 Channels” HackEEG:
https://docs.starcat.io/hackeeg-crowdsupply/user/software-installation-32-channel.html
- But we never have seen more than the prototype device which is now abandoned.
PEEG
- is limited to 16 channels
- is new on the market
- There are rumours that it has issues with EEG noise
- There are rumours that is an abandoned prototype of Neurosity CEO A.J Keller.
- PEEG has not yet proofen be independent testers to be free from EEG noise.
- We only know of demonstrations of PiEEG which use powerful SBCs, which is not optimal, and a demonstration which moves a robot toy very basically.
32 channel EEG devices based on the TI ADS1299 demo kit
- Indonesian scientist have built a 32 channel EEG device from the ADS1299 demo kit
- It uses a single board computer as well, for instance a Raspberry Pi.
- But devices based on this architecture neve came onto the market.
BEATS
- Chinese developers have built the «BEATS» device with 32 channels
- It is only a demo device which did not come onto the market.
- It uses a Raspberry Pi 4 and four ADS1299 shields in daisy chain mode, simililar to the device from the Indonesian scientists.
- Is the cause that it is not on the market EEG noise?
- Is the cause for this that the Raspberry Pi 4 causes too much noise?
SMARTING mobi
- Has 24 bit A/D, this is the same rate as the ADS1299 has.
- we did not find from which EEG chips it is built from, but it could be ADS1299.
- According to the specifications the smarting mobi device with 24 channels could be build up from three ADS1299 chips which gives 24 channels.
- This device is on the market.
mBrainTrain
- Is sold on the same webstie as SMARTING mobi
- It could be build from the ADS1299 as well, but we do not have any information about it.
- They offer 32 – 64 channels
- This device is on the market
- According to a paper published in NeuroImage,
Volume 269, 1 April 2023, 119774 - Wireless EEG: A survey of systems and studies
10,1 percent of wireless EEG users use devices from mBrainTrain, but the leader is Emotiv with 24.6 percent:
https://www.sciencedirect.com/science/article/pii/S1053811922008953
So the only option we have for instance for a 32 channel EEG device based on the ADS1299 is to buy an expensive device from mBrainTrain. There are papers for the BEATS 32 ch. EEG device, but if the BEATS would be competitive, would it not be on the market and compete with devices from mBrainTrain?
What says Peter Gamma from www.petergamma.org to it?:
After studying devices based on the ADS1299 EEG chip from Texas Instruments for many years, he came to the conlusion that this chip is suitable for building EEG devices for medical and research grade applications up to a large number of channels, and he could in principle build such devices by himself. But you need a lot of soft- and hardware expertise for it, and many experts seem already to have given up this topic. And Peter is a physiologist, and not a EEG device builder.
He therefore hopes that the NeuroKit2 developers update their software very soon:
to make out of it a physiological multi-sensor platform based on InfluxDB, which can be used with 32 TGAM EEG modules to build a 32 channel EEG device for 700 USD with it.