Ultracortex first readings garbled
Hello all!
My name is Juris and I`ve recently picked up on this new amazing thing called OpenBCI. I study game development and I have some ideas what to do with it, but I`ll keep the long story for my website.
First I just got myself a 16-channel set of all the modules and things. I assembled it all, following instructions/reading forums and it seemed to work fine, but I did not like the inconsistency and the messiness of setting it all up so I found the Ultracortex Mark 3 and printed it out at my university and got all the parts and assembled that too, carefully following the instructions and all, however, I have run into some issues when testing it and I`m not sure why is it acting the way it is. I`m running it on the OpenBCI GUI compiled by Processing 2.2. I`ll list the issues along with my theories as follows:
In general there seems to be a massive amounts of noise happening way above what the electrodes should read in the microvoltage readings in most of the electrodes (in hundreds or railing), while it was in the range of 15-20umv at three out of the eight electrodes. At first I tried separating out the connected electrodes to test them one by one (using the srb and bias on my left and right earlobe respectively), but then I thought:
1) What if I don`t connect the other pins, can there be some noise happening just because of that? As in do all the 'open' pins and connections on the board, that are not connected to anything, create any noise? (Many tutorials and pictures have these kind of black plastic blocks plugged onto the empty pins and into the unused notches on the board and I`m wondering if I should be doing that too.)
2) As I was assembling the electrodes following the Ultracortex instructions (https://github.com/OpenBCI/Ultracortex/tree/master/Mark_3) I discovered that the springs I ordered are slightly narrower than the ones used in the tutorial so the only option I could think of was to attach it like so (see picture). I am very worried that the spring (which is made out of conductive metal) is touching the hex-nut that holds the wire thus when the spring contracts it could cripple the reading. If so, could I somehow isolate the whole conductive part with the two hex-nuts and the wire under a layer of superglue, for example, and then prop the spring against it? Would that work? Would superglue even isolate it? Or is my only option to get springs wide enough so they don't touch the conductive part? Here's a picture of what it looks like now: You can see how I've tightened the spring around the second hex-nut that holds the wire and just barely fit it around the very end of the plastic around the wire. It actually holds up very well for the spring to be stable at least. However, this being my electrode, only five out of eight showed completely garbled umv reading in my first runs (100< and railing). The electrode in the picture does not show, but some of my electrodes also have the wire, that is fixed in-between the hex-nuts, slightly split and not so nicely tied up in a twist. Could this also be a potential issue?
3) What about the wire management? I was reading a book called Brain Computer Interfaces: Principle and Practice and it talked about how pretty much any smallest wire movement (amongst other minute details) can impede on the reading. Is it safe to assume that OpenBCI has some kind of measures that prevent such extremely sensitive behaviour? Like should I measure and cut the wires on my kit to the centimeter so that they don't have to be bent and tied down? (A bent wire technically folds over its own electromagnetic field - this could be an issue, no?) This is what my Ultracortex looks like. Does it seem alright to an experienced eye?:
4) Would the test with an open-ended wire without an electrode between both of the reference aligator clips work to show a close to zero microvoltage reading? And could I safely assume that my assembled electrodes are at fault if this test passes but the actual electrode readings are garbled?
5) Here is a picture of the reading test: As you can see, only five electrodes show the totally garbled up reading. Or are the readings in the +-20umv range also unusable garbage? Does this prove that some of my electrodes function ok eventhough all of them have the spring component touching the conductive section?
I am not too familiar with the hardware electronics stuff and assembling things like these and I would be eternally thankful if someone more experienced could tell me which of my concerns are valid and which are not, or some other hints to what I could do.
I apologize for the long post. Please don't hesitate to ask me to clarify something if needed.
Comments
Can you clarify this point: originally you had the 8 and 16 channel kit with gold electrodes and paste. And at that time everything worked correctly. Is that right? That would tend to say that all your current problems are related to the headset and dry electrodes, and not the electronics.
It does look like you've got huge mains noise. Are you sure your Bias (Ground) clip and lead are working well. What happens when you swap that with the Reference clip lead. You didnt mention what ear clips you are using. You should not mix metals with the electrodes. Use Ag-AgCl ear clips with the Ag-AgCl combs.
The wires do not have to be cut exactly, see the thread on Mark III/IV feedback. I'm not sure how your springs could be causing this.
William
You MUST set your notch filter to correspond to your local mains frequency, 60 hz here in the US, 50 hz for you. With your notch set at 60 hz, you are getting the huge mains noise at 50 hz. Press the button and see what happens. It should reduce the noise.
A 'notch' filter removes noise at the frequency specified. As opposed to a 'bandpass' filter that passes frequencies in that range.
By mixing metals I meant, don't use gold cups and silver chloride electrodes at the same time. While this can work sometimes, it can also produce what is called a 'battery effect', from the two metal types being connected together through an electrolyte. If you use the ultracortex instructions, you'll be all sliver chloride.