Shielding to minimise noise for OpenBCI EEG electrodes

Tags: #<Tag:0x00007f628d2a6e20> #<Tag:0x00007f628d2a6ce0>


Most of the setups for passive electrodes I’ve seen use unshielded cables, and it appears likely that the longer the cable, the more prone to electromagnetic interference such as 50/60Hz mains it will be.

Does anyone have experience with shielding, particularly in the context of a OpenBCI EEG setup?
I’m thinking of using coaxial cables for electrodes or some kind of foil “hat”.
Either way, I’m uncertain if this would distort the signal by capacitance as well as which “ground” to connect it to. I’m guessing Reference or Bias, but I really don’t know.

The best I found on a search so far was this: Noise reduction strategies in electrophysiology but I’m having trouble interpreting it in practical / OpenBCI terms.

Any ideas or experiences on this?


Thanks for that great link on the electronics Stackexchange.

Except for very high end EEG systems or those that use active electrodes, shielding is not commonly used. The best strategy is to just minimize your cable length and cable motion. The Ultracortex headset is one approach.

Another way to go if you are using a velcro headset or cups/paste, is to clip the OpenBCI onto the front chest of your shirt. With the closeness you can then use short cables such as 3 or 4 foot in length. Pocket-Neurobics also uses this clip on idea with some of their amps.

The Ground or Bias electrode on OpenBCI and many amps provides a reverse polarity mains noise cancellation signal.



Thank you for your input William!

I think I’ll get the FRI 19 channel headband. Based on yours and others advice I would likely strap the BCI board to the back of the head and use the 18 inch/45 cm cables along with a pair of ear clips. That would minimise the cable length as well as potential for movement. Only the battery would be worn in pocket which likely wouldn’t affect noise.

If I understand you correctly, if I were to apply shielding of some sort even though it’s not commonly used, I should connect it to OpenBCI Bias/Ground.
I would then apply an experimental shield and see if it makes any difference to a baseline noise measurement.

Is there any best practice with the OpenBCI for unused electrodes, should they be tied to reference or is it unnecessary regarding noise?

I’m also not quite sure how to measure just the noise component, without a body attached. I found this on measuring Self-Noise of OpenBCI. I want to measure the noise with the cables attached to the headband to measure the noise floor with and without shielding. Should I leave the electrodes open, short them to reference and ground, or with some series resistor to replace the body?



re: unused electrodes. These have no effect on noise.

re: noise measurement with your shield tests. I would measure with the headset on your head. Just eyeball the FFT graph with the shield connected or not. If indeed the shield is improving signal to noise ratio, then you might expect the entire FFT graph to drop a bit, or in some areas.

In general shielded EEG cables are a luxury. They are not seen in the field very often. The Nexus EEG amps use them.


Wonderful, that clarifies all my questions. I’ll simply try makeshift shielding, eyeball the result, and if there is a difference I’ll consider taking it further with DIY foil or DIY coaxial cabling.

I’ll report back when and if I reach any results.
Thanks for all your help!


Aluminum foil shielding would be problematic because of the poor conductivity at joints or solder attachments. Your FRI leads already have touchproof and clamp connectors, so adapting to coax would be unpleasant.

This is an alternative, you may find smaller diameter online,