openeeg research

openeeg research

Table of Contents

Documentation, research and code around the openeeg project


  • modeeg PCB/printed circuit boards/parts



    About the design: http://openeeg.sourceforge.net/doc/modeeg/modeeg_design.html

    • 1] Two analogue and one digital board ordered from Olimex giving four channels.
    • 2] Parts ordered and assembled following BOM:




    [which also includes instructions for testing, troubleshooting and trimming of amplifier boards]

    See also http://openeeg.sourceforge.net/buildeeg/



    • 3] Construct 2x17pin interconnect for digital and analogue boards (using an old ATA

    HD cable)

    • 4] Analogue boards were placed in a metal case with interconnect to

    digital board and shielded (RG58) leads breaking out to further cinch connectors on the plastic case.


  • testing

    Boards were tested and trimmed following instructions at:


    Several of the INA114 op-amps were faulty, and the odd intermittent fault surfaced.

  • programming

    :cvs -z3 -d:pserver:anonymous@openeeg.cvs.sourceforge.net:/cvsroot/openeeg co -P fw

    Addition of macros to source code from:



    :avrdude -c usbasp -p m8 -U lfuse:w:0xFF:m -U hfuse:w:0xDF:m -U flash:w:atmega8modeeg-p2.hex

  • DIY electrodes


    Headband and adapters were based on: http://openeeg.sourceforge.net/doc/gallery/joe/index.html

    and sponge electrodes roughly on: http://openeeg.sourceforge.net/doc/gallery/nelo/index.html

    using silver electrodes and salt water solution in the cleansing sponges.




  • Placement of electrodes and grounding
      The "ground" used for (cable) screening purposes and connecting to the client (if you don't use the DRL) is the virtual ground called VGND on the schematic (it is obtained from the output of opamp 201A, pin 1).   Connecting to the power ground will not work and is dangerous from an electrical isolation point of view.  (Meaning your brain can get fried if a catastrophic failure would occur as a result of a lighting strike into you power grid as an example.)   About 50% of commercial EEG amplifiers used for NFB and most, if not all, clinical EEG amps ( those that have 19+ channels) use the medical safety DIN connectors.  As these connectors only have a single pole, equipment using these generally do not used screened cables.  In fact, the only amplifier that I know off that uses DIN connectors WITH screening is the MindMaster which Joerg helped to develop.  The MindMaster uses 2 DINs for each electrode.  One DIN for the signal and a second DIN for the screen.  The advantage with this setup is that you can buy standard ready made cables (unscreened) and use those.  If you run into severe 50/60Hz noise problems you can then buy/make the more expensive screened cables.   Many-channel EEG amps such as BioSemi that don't use DINs but a sort of ribbon connector are, as far as I am aware, not medical grade equipment and are intended for research labs, but not hospitals.  There is simply no way that equipment that uses such connectors can pass the stringent medical grade requirements.   If you don't want to use DIN connectors I would recommend RCA/phono connectors.  They are cheap, plentiful and you can implement screening by default.  If you do use screening, connect the screen to VGND or the pad labeled 'PAD205'.   Also remember that the screen of the actual cable will only be connected on the amplifier side.  The far side (electrode side) will be left unconnected.   If you only use one channel, you should connect your third (reference) electrode not to DRL but rather to VGND (this will however increase 50/60Hz noise).  Or, if you do use DRL, connect CH2+ and CH2- to VGND or DRL.    The metal box I would connect to AGND because of the lattter's low impedance.  Just make sure that the metal box is very, very far removed from any components on the non-isolated side (such as IC107, IC103/4 and IC106) and the RS232 and DC jack connectors.  You want a physical distance of at least 10mm to prevent flash-over due to a lighting strike or whatever.

    [Stefan Jung from openeeg-list]

    Each channel has two electrodes. Each channel has + and - inputs (electrodes), and calculates the difference of the input voltages. This cancels out or attenuates common mode signals like mains hum that (ideally) have the same signal strength on + and - inputs.

    With 2 channels 5 electrodes (including DRL) are required (or at least 4 when 2 channels share one reference (-) electrode)

    DRL is an output that acts as an active earth and suppresses mains hum (at least in theory) even more than a simple connection to VGND. Shielding You will need to use shielded cabling for all passive electrodes. The cables from the electrode store are not suitable. This is necessary to get the DRL working effectively. Andreas said:

    First, any cheap shielded cable should work. Not too cheap though. Very low cost audio cabling can be worthless, that is have very little shielding.

    Assume you want a 1.2 m electrode cable (longer than 1.8 m is not recommended). Have both leads for one EEG channel run in the same cable, sharing the same shield, for about 1 meter. Then have the cable branch into two leads the last 20 cm (or longer if you have a big head. ;-). The branches are also shielded of course, up to the electrode.

    In short, it should look like a Y.

    By doing this, you reduce the capacitive loading on the electrode leads, and get more managable cabling too.

    Just remember to only attach the cable shield to the amplifier board, and not the electrode itself.

    from: http://openeeg.sourceforge.net/doc/hw/electrodes/passive/

  • DIY electrode paste

    "About electrode paste: I can't help you there. I am using essentially the same thing: home made starch/gelatine gel with salt water. Cheap, but without any adhesive properties."

    from: http://openeeg.sourceforge.net/doc/hw/electrodes/passive/



  • extension



Author: m <m@1010.co.uk>

Date: 2010-01-13 16:56:46 GMT

HTML generated by org-mode 6.31trans in emacs 23