DIY headband, all channels railed, known good electrodes
Shawn
Chicago
Hi,
I am trying to use the snap electrodes to collect EEG data, as shown in the pic below. But I got all channels railed after started. I ran the impedance test and pretty much all electrodes are in a good range. I am using Cyton board for this project. Does anyone have any clue why this happened?
Many thanks!
Comments
Shawn, hi.
'Railed' generally means that your signals are peaking ABOVE the maximum microvolts level set in the upper right on your time scale graph. Try setting a larger maximum. Most normal EEG is below 100 uV, so if you are getting huge values like this it may be related to electromagnetic noise in your vicinity. Try moving to other areas away from extension cords, power supplies, lighting fixtures, conduits in walls ceilings floors, etc.
There is also a whole thread on resolving RAILED when using Ultracortex. Found using the Google Search button in upper right of the Forum:
https://www.google.com/search?as_q=railed&as_sitesearch=openbci.com
Finally, with the dry electrodes and headbands, pressure on the electrode has a strong effect on the quality of contact with the scalp. That is why the velcro bands are used, with user adjusted tension. I assume your custom headband has such tension / pressure adjustments. The velcro does NOT flex, so tension applied directly scales to electrode pressure. If your band is elastic, that may not be the case. The velcro material is easily available, and can be hole punched:
https://www.amazon.com/VELCRO-Brand-ONE-WRAP-Double-Sided-Multi-Purpose/dp/B000078CUB/
William
Thank you for your reply, William.
I tried with a larger range, as well as "auto", they are still railed. Any other possible reasons? Or any other check I should make?
The impedance is good means the electrodes are well functioning, right?
I also did the bridging for channel 1, bias, and reference. Is this the right way to do it as in the picture? and the result?
Did you read the other pages on fixing 'railed'?
https://openbci.com/forum/index.php?p=/discussion/1481/how-to-become-de-railed
If you connect together a channel electrode, reference and Bias/ground, the result should be close to zero.
Did you try moving your equipment to an electrically more 'quiet' area? Did you try adjusting band tension as recommended? If your band is rubber-stretchy, then that may not have the desired effect. The velcro one-wrap works the best as tension translates into electrode pressure.
Certain hair products, conditioners, rinses, etc., do not play well with EEG. Generally speaking, for EEG recordings it is advised to shampoo hair so scalp is clean, but not to use any after-products.
Yes, I have read it. Actually, I have another set of Mark IV headset and electrodes which works correctly. The reason I customize it is because I need it to be compatible with another Hololens device.
And I am pretty sure I have enough pressure for the electrode.
And when I bridged them together, I got a result 0.13 uVrms as shown in the background in the pic. Is that normal?
Can you please help confirm if the electrodes are well functioning when I got the impedance test 30-90 kΩ?
Many thanks!
If your Ultracortex is working, and your band system is not, then clearly you have some problems with the band. I would first suspect pressure on the electrodes is intermittent. Highly suggest just getting the One-Wrap and a hand operated hole punch.
https://www.amazon.com/s?k=leather+hole+puncher
Yes, as we stated, shorting all three electrodes should produce close to zero uV.
Impedance alone cannot guarantee quality of EEG, but your values less than 100K should be fine.
One further advantage of the velcro one-wrap, is that you can reach other areas of the 10-20 map. This is done by having your base band in place around the circumference of the head. You can then add 'cross' bands that go over the top of your head. Provided that the tension on the base band is high enough to prevent the cross bands from pulling the base band upwards. A refinement adds a "chin strap" under the chin, if you need to exert downward force on the base band.
See photos on this page:
https://fri-fl-shop.com/collections/headbands-amp-fasteners
So since you already have your connectors and electrodes, you just need a roll of the one-wrap. Link above is 12 feet.
Thanks for the suggestion. I did use my hand to hold one sensor and make sure I put enough pressure on it to my scalp, and I tried multiple times with different areas on my head, didn't work.
My Ultracortex is using different types of electrodes (spikey comb electrodes), that's why I am wondering if it's an electrode problem. Because it's not just one or two channels not working, all of them not working with this type of electrode. Do you have any other suggestions to help with finding and fixing the problem?
Many thanks!
Since the situation is happening with virtually ALL your channels, I would expect an issue that is in common with ALL channels. Such as your reference connection, lead, connectors, electrode, contact with skin. Or Bias.
The 5mm combs DO wear faster than the spikey combs. If you see whitish plastic or silver beneath the tips, that means your silver-chloride comb tips are worn out. Silver chloride electrodes are a uniform dull grey matte color.
If you read the earlier links, they also suggest slight rocking of the combs to ensure the hair is cleared out of the way.
You never mentioned what you are using for reference and bias electrodes. You cannot 'mix' electrode metal types, such as using gold, tin or silver ear clip electrodes. Mixed metal types cause 'galvanic' battery effects, misleading the actual voltages being measured.
Since your Ultracortex works fine, there is certainly something wrong with your headband setup.
I use these as reference and bias:
https://shop.openbci.com/products/earclip-electrode
And these electrodes:
https://shop.openbci.com/products/eeg-snap-electrodes
@wjcroft, thank you for your help. I just found that if I turn off BIAS in the panel, the issue could be fixed. (Please ignore Channel 4 in the pic)
The problem is how well could I trust the data without BIAS, and any other steps that need to be done when I play around with the data afterward?
I'm recalling now you had this earlier issue with the Bias:
https://openbci.com/forum/index.php?p=/discussion/3543/the-channel-of-c3-electrode-railed-if-connect-the-bias-channel
This really sounds weird. But it's fine to not use the "Bias injection". You can still connect the Bias electrode, and just switch off the 'Bias' counter-mains feature using the settings panel. Doing that will only slightly increase your mains (60 Hz) artifact. But since you are likely already using a notch filter at 60 Hz, not really that big an issue.
Bias has two functions: centers the differential amplifiers, and does counter-mains injection. The switch in the settings just turns off the injection. The amp centering still happens.
@wjcroft
You got me there. hahh
But last time was just one channel and has been proved it was the worn electrode problem.
This time all channels railed with BIAS, they are all 100% new electrodes. It's so hard to believe all new electrodes are malfunctioning.
One more question: does turning off BIAS for channels infect other features in OpenBCI GUI? like focus, etc?
I continue to believe there is something 'unusual' about your headband that is causing this, since the Ultracortex works fine. Is it possible that your Bias cable, connectors, electrode, etc., could be subtly "out of tune" with normal EEG? You can do things like swapping components to test this, using an ohmmeter, etc.
re:Bias 'infection'.
I already explained this, it will only have a slight effect on the 60 Hz mains noise. Which is already being addressed by your notch filter.
Hi William @wjcroft, I am wondering if is there any place I can get more components like the one in the pic (not the comb electrodes)? I would like to make some spare ones to back up my experiments. Thank you!
Shawn, hi.
Your picture shows the Ultracortex electrode holder with a comb electrode mounted. These are included in the various versions shown in the Shop,
https://shop.openbci.com/products/ultracortex-mark-iv
However if you just want some spares, email to (contact at openbci.com), customer support and ask how to order / pricing.
William
Thank you for your timely response, the blue part in this pic is self-printable, right? But I can't find where to get the wire and metal part next to it in the shop.
Email to customer support.
I believe the Ultracortex IV electrode holder and what it slides in, are all injection molded. Because they needed tighter / smoother tolerances than the 3D printing could produce. There have been some Forum threads in the past, where the cable you show, has a break between the wire and the nut that crimps it.
So these folks were also looking for replacement parts. (The crimp nut, wire, Dupont connector assembly.) It was suggested at that time that spares be listed in the Shop. That apparently never happened. When you communicate with them, you might mention this.
re: the blue plastic electrode holder piece: They may also stock these, although this is very unlikely to malfunction.
Okay, Thank you so much.