Ultracortex Mark III Feedback & Mark IV Priorities

Conor RussomannoConor Russomanno
edited November 2015 in Headware
Hi everyone, 

I want to start this discussion regarding:
  • Issues w/ printing & assembling of the Mark III
  • Feedback/suggestions on assembly guide & the Mark III design
  • Design priorities for Mark IV
  • Other 3D-printed headset conversation topics
This thread will heavily influence the design priorities of the Ultracortex Mark IV, from our ongoing Kickstarter campaign.

Comments

  • wjcroftwjcroft Mount Shasta, CA
    edited November 2015
    Conor, thanks.

    The current assembly guide suggests choosing a single 10-20 electrode site placement layout, then custom cutting the electrode leads to match. If the user then needs to change the layout (say for another BCI or neurofeedback requirement) -- it's possible that some of the shorter electrode leads may no longer reach their new 10-20 sites.

    Some mention of this in the guide might be helpful. Possible solutions may be to leave the leads longer, then use twist ties or tie wraps to coil (or zig zags) to take up the slack. Another approach may be to have some extension leads handy if faced with a short lead. Adafruit has these available in 3", 6" and 12" lengths. The links at the end of that page also show other jumper leads sets available. These could be handy if the user does not want to sacrifice their cup electrodes.

    William

  • Conor RussomannoConor Russomanno
    Great suggestions, William. I think both scenarios you've detailed are good solutions. The current Mark III kit comes with enough electrodes (21) to create an electrode unit for each site. This is a tad wasteful, however, since you can only target a maximum of 16 of the 21 nodes, and many people are working with just 8 at this point. 

    For the Mark IV kit we will likely ship the kit with only enough electrodes to run the 8-channel setup, and we will offer a cheap, add-on kit for targeting more locations. The standoff nodes (without an electrode) will be redesigned with a semi-spherical, soft tip and roughly a dosen of these will come with each Mark IV kit.
  • coreygocoreygo
    I left the wiring on my Mark III practically full length and zip tied them neatly to reduce movement and keep things relatively clean. I wanted to be able to just cut the zip ties and move stuff around if necessary. Pretty much what wjcroft suggested. I did finish and install all locations but left the unused locations somewhat unscrewed and tidy but disconnected for when I had the remaining 8 channels.

    Instead of using sandpaper and sanding everything by hand I found two attachments for my Dremel which made that job super quick and easy.

    I also used paper tags to label each electrode. My only piece of feedback would be to modify/enhance the charging method for the LiPoly battery from Adafruit. I'm a little worried that the plastic will wear down so I'm using 2 screws instead of 4 to hold the board in place. I'm not sure what they're called but I thought perhaps those plastic pieces that are used to hold heatsinks in place might work well.

    Regards,
    Corey
  • wjcroftwjcroft Mount Shasta, CA
    Here's an idea about how you could streamline the electrode holder assembly, eliminating the gluing step. The bolt / nut attachment in the new electrode holders would also allow replacement or changing of the silver chloride combs / cones. (The current glued electrode holder does not allow easy access to the nuts after gluing.)

    The revised idea is to just print the holders as one single piece. In other words the hollow tube portion will be longer. How to attach the wire and nut? Just make the bolt longer, so that it almost projects out the end of the tube.

    Then to attach the wire, instead of using double nuts, do this: Insert the bolt through the comb, then into the holder. Place a lockwasher on the threaded bolt end, and let it fall towards the comb end. (Be sure to use the toothed washers like this rather than those with only a single split in the ring. The multiple teeth grab better.) Now take your wire and form a twisted loop on it's end. This can be done by using a convenient rod or screwdriver, a bit larger than the thread diameter of the bolt. This "wire loop forming" tool could also be a simple rod that you 3D print; again, slightly larger than the bolt diameter so that the loop will slip down over the bolt and not catch on the threads.

    Loop the stripped wire end around the rod and twist several times, the insulated and stripped ends. Such that the loop now is maintained in the wire. Cut off the excess stripped portion so that only several turns of wire are needed to hold the loop in place.

    Bend the loop so it is now at 90 degrees to the insulated portion. Slip the loop over the bolt end. Slide it down the bolt end. Add another lockwasher, then the final nut.

    To tighten the nut, you can use something like one of these handy nut starter tools. That could either be purchased in bulk, or printed.


    This particular tool is made of flexible plastic such that it grips the nut and also has a hole through which the bolt can pass. But you could print something similar. One end fits #2 nuts, the other #4, #6.

    Thread the nut down to where it meets the end of the holder, then tightening from the bolt end. Once the initial tightness is there, the lockwashers should hold it firm while the comb bolt is torqued to it's final tightness.

    I should have put a picture here to clarify, but I think this text should describe the process. Let me know if you have any questions.

    William
  • Conor RussomannoConor Russomanno

    A comment from John Pellman on the OpenBCI/Ultracortex Github Repo: "

    -- High Density EEG --

    This might be more of a suggestion for a future project than an improvement on the present iteration of Ultracortex, but I thought I'd throw it out there anyways.

    A small, but growing, number of researchers are scanning with 32, 64, 128 and 256 electrode systems in order to employ a technique called source localization. Source localization, as its name implies, allows you to pinpoint where EEG signals are coming from in a way that gives spatial resolution reminiscent of fMRI (though not comparable; fMRI voxel sizes can be as low as 1 mm isotropic, while EEG voxel sizes tend to be around 5 mm isotropic). This technique, though not without controversy (what scientific methodology isn't?) is something I would be interested in seeing become more prevalent, as it seems to me that it would be much more cost effective to gather large datasets with EEG than with fMRI.

    There are some high density nets on the market today, and while they're cheaper than fMRI, they can still be rather pricey (upwards of a thousand dollars per net). In a time of dwindling scientific budgets, these costs can be prohibitive to some labs that otherwise might have wanted to try using this technique. In light of this, I think that it would be neat if there were an open source alternative that labs could try out (even if it weren't necessarily as fully-featured as the nets that are out right now).

    Here are a couple links for you to check out concerning source localization if your interested:
    http://cdasr.mclean.harvard.edu/content/publications/LATN/BookChapters/Pizzagalli_HandbookPhysiology07.pdf
    http://neuroimage.usc.edu/brainstorm/Tutorials#Background_readings

    "

  • wjcroftwjcroft Mount Shasta, CA
    Source localization is also being done with the 19 channels of the 10-20 system. Such as with the Brainmaster Discovery amplifier hardware and Avatar software. So although high channel counts increase voxel resolution, as you've noted there is a considerable cost factor. Many clinics are already employing 19 channel 3D neurofeedback using variants of the LORETA software. I'm looking forward to the further evolvement of these tools.

  • wjcroftwjcroft Mount Shasta, CA
    Regarding ease of recharging the LiPo cell and reducing wear on the housing screw threads & connectors,

    http://openbci.com/forum/index.php?p=/discussion/592/connector-housing-wear-after-many-rechargings

  • wjcroftwjcroft Mount Shasta, CA
    edited February 2016
    Here's some links on the type of tDCS hydrophilic foam used by the clinical tDCS companies, such as Soterix. The advantage of this foam material over "sponge" is the microscopic pore size. This translates into better comfort as the course holes in sponge material create higher current gradients, thus potential for uncomfortable sensations as the current ramps up.

    McKesson 61-84344.  Here is one site carrying them, with photo.  I got a pack of 10, which can be cut into many squares or disks.

    Amazon also carries, as well as other online stores:

    https://www.google.com/search?q=mckesson+61-84344

    As far as the saline solution, the more dilute, the less harsh the sensation.  As these guys documented.  I'm using about a 60 mMole solution.  Just a tiny tiny amount of KCl in 100 ml of water.

    http://dx.doi.org/10.1016/j.clinph.2007.01.010

    William

  • Conor RussomannoConor Russomanno
    Thanks for the info, William. This is very helpful.

    Conor
  • wjcroftwjcroft Mount Shasta, CA
    edited March 2016
    Softening FRI combs that may feel too pointy:

    I assume you guys are considering the flexible Ninjaflex filament for some applications such as bumpers or couplings. Another possibility would be as "softeners" for the FRI combs, in those cases where certain users feel them on the scalp as too "pointy" during long sessions.

    So I'm envisioning little Ninjaflex 'pads' that have an inverse hole pattern to match what is on the combs. These can friction-fit over the combs and thus offer some cushioning effect. They could be printed in various thicknesses or "textures" (surface patterns) that could provide various degrees of softening of the perception of the FRI comb pins on the scalp skin.

    If you've seen this stuff you know that it can be made more or less "squishy" by how many voids and cavities are inside the interior. Thinner walls and more voids create an object that is softer than one that is printed as a solid block. Softening of the Ninjaflex could also be achieved by using a surface texture such as ridges or comb teeth.

    William

  • wjcroftwjcroft Mount Shasta, CA
    On the dry sensor thread, below link to a new post on graphene based conductive filament. VERY low resistance, .6 ohms-cm. And the printed objects are also stronger / smoother than other ABS / PLA filaments.

    http://openbci.com/forum/index.php?p=/discussion/comment/3821/#Comment_3821

    Joel @biomurph and @Conor did some earlier test print combs from conductive filament. But the filament material was of too high resistance. (It required gel injection to get a good connection.) This new graphene filament may solve that. And since it is also stronger / smoother, may allow comb shapes that are analogous to the splayed design of the Cognionics combs. That is, comb tines that bend / splay out a bit as they contact the scalp, thus providing some adaptive qualities to scalp surface variations.

    William

Sign In or Register to comment.

© Vanilla Keystone Theme 2026