noise spikes seen with Ganglion

Radha_openbciRadha_openbci United Kingdom
edited November 2020 in Ganglion
I have been using Open BCI Ganglion for quite a long time now(7 months). I purchased a new board and started using it a month ago. I am using Gold cup electrodes. Since a week it has started giving more amplifier saturations and sometimes it samples very slow. Typically, now the data can look like:


The first 3 are raw from the time it was sampling slow. The last one is filtered, and I am referring to the double spike on the left that comes frequently  now. I am using bled112 dongle. What can be the issue  here?

Comments

  • wjcroftwjcroft Mount Shasta, CA
    Radha, your 1st three images appear to show lots of mains frequency noise on primarily the CP3 electrode, but other channels are also picking it up.

    Have you tried swapping your electrodes around or using new electrodes? If the gold plate gets worn or chipped, it can distort the signal quality. 

    Cleaning the paste from gold cups in a particular way, can prolong life -- by reducing any rubbing.


    Regards, 

    William

  • Radha_openbciRadha_openbci United Kingdom
    edited July 2019
    Well to troubleshoot the problem, we tried various things:
    1. Changing electrodes doesn't help
    2. Streaming LSL so could be a memory issue causing packets to loose, but NO, I have 32 GB RAM
    3. Read on another forum that it can be a charge accumulation issue since our encasing is a plastic box. Our encasing is plastic but 1 side is metal (near the pins). We keep it on a plastic chair when we record. When the chair is touched the peaks definitely occur, but not when it is frequently touched. Otherwise peaks occur randomly. So we removed the box and checked, it still happens. 
    4. Connected the G pin to the metal but no luck
    5. Tried in 4 different environments, still happens

    Do not know what else to try.
  • wjcroftwjcroft Mount Shasta, CA
    In rare circumstances, the flash memory in the processor (Simblee), has a corruption / bits altered. Which can affect operation. You can reflash using the instructions here,


    The images you showed look primarily like mains noise. What types of filters are using using? Notch / bandpass?

    William

  • Radha_openbciRadha_openbci United Kingdom
    Both notch 50 Hz and band pass 2-35 . But the raw signals reflect it and despite the 50Hz notch on Open BCI GUI, it is evident always. I know the notch there is only for viewing .

    Radha
  • Radha_openbciRadha_openbci United Kingdom
    Now at least it is not sampling slow frequently, but the dual peaked high amplitude is my concern that happens abrubtly.
  • wjcroftwjcroft Mount Shasta, CA
    Does the time series graph in the OpenBCI_GUI (with notch and bandpass turned on), look similar to what you are seeing in Matlab / EEGLAB? Or does it look cleaner in the GUI? If it looks ok in the GUI but not Matlab, that could indicate something is amiss with your filtering in Matlab.

    Does the FFT graph in the GUI (without notch) show a large frequency component near your mains frequency? Is it EXACTLY 50 or 60 hertz, or some multiple (called harmonic)? Harmonics can also be subharmonics.

  • Radha_openbciRadha_openbci United Kingdom
    I can share here the raw files of what I am referring to. No in the FFT everything goes higher during those dual peaks, actually sways up and then comes back to normal. In the GUI I don't switch on the band pass. Yes, it is fairly similar.
  • wjcroftwjcroft Mount Shasta, CA
    Please share the GUI screen, WITH bandpass and notch turned on.
  • Radha_openbciRadha_openbci United Kingdom
    Now I remember the BP is always on. Nevertheless here it is:
  • wjcroftwjcroft Mount Shasta, CA
    OK, so with notch filter turned on, the GUI shows ZERO mains noise. Yet still shows spikes occurring periodically. In a previous post you said:

    "When the chair is touched the peaks definitely occur, but not when it is frequently touched. Otherwise peaks occur randomly."

    So this tends to confirm that the spikes are related to some type of electrostatic buildup.

    Your Ganglion when originally received worked fine for at least a month, according to your first post on this thread. Then last week the spikes began. What did you change? Can you replicate your original circumstances?

    I'm not sure where your spikes are coming from, it could be related to clothing, hair products, etc. In clinical settings it is advised to wash the hair before EEG sessions, but NOT to use any hair conditioners or other products. These can cause unusual distortions.

    Regards,

    William

  • Radha_openbciRadha_openbci United Kingdom
    Hi!

    Thanks for the feedback. But the same happens in now 4 different environments on all people. I don't know why the electrostatic buildup would happen even after opening the encasing.
    Before a month i was using another Ganglion and it seemed to work fine for many months. I will post here if i figure out something.

    Radha
  • wjcroftwjcroft Mount Shasta, CA
    Can you please run some tests with the old Ganglion to see if the issue disappears with your current experiment setup? That would then confirm that something has gone wrong with your new board. If that is the case then I suggest reflashing. 

    Where is the Ganglion Ground pin (D_G) connected on the head? Don't connect anything to GNDA. Some clinical EEG connects Ground to a site on the midline, for example AFz with Electro-cap. Or any unused midline such as Pz CPz Cz CFz, etc.
  • Radha_openbciRadha_openbci United Kingdom
    I connect it on forehead. Unfortunately it is not possible to test both simultaneously now as both are in different countries. I can test both in a month when i have them together.
  • wjcroftwjcroft Mount Shasta, CA

    @Radha_openbci, hi.

    Here is one of the Cyton graphs you posted on the NeuroTechX #tutorials channel.

    Was your previous Ganglion noise similar in character to this Cyton noise?

    William

  • Radha_openbciRadha_openbci United Kingdom
    edited November 2020

    Hi
    This is from Ganglion too and I am getting these bursts for the first time if I recall correctly. Now I have both the open BCIs I can check signals from both in the same environment. I use one if the other is producing noises. Whats the best way to compare?

  • wjcroftwjcroft Mount Shasta, CA

    So if BOTH Ganglion and Cyton are showing similar noise patterns, does that not tend to eliminate a possible issue with the Ganglion or Cyton mainboards? But instead points at either an issue with local environment electromagnetic noise, pops in the electrodes, cables, etc.

    Electrode pops can happen for a number of reasons.

    https://www.google.com/search?q=eeg+electrode+pop

    What electrodes and paste / gel are you using now? Same as with both Ganglion and Cyton?

    William

  • Radha_openbciRadha_openbci United Kingdom

    Just to be clear, I do not have any Cyton. These ae 6 channels because the last two are bipolar derivations from 1-2 and 3-4

  • Radha_openbciRadha_openbci United Kingdom

    I don't think its the local environment because the one that had initial errors are still behaving that way in another location. Now I am using brand new Ag/AgCl electrodes with Ten20 gel, same with both.

  • wjcroftwjcroft Mount Shasta, CA

    OK, my misunderstanding of the six channels shown. Did you say previously that the same type of noise was showing up with multiple Ganglion boards? This again, seems to eliminate the mainboard as the issue, and points to something else going on.

  • wjcroftwjcroft Mount Shasta, CA

    This page has some suggestion on the 'electrode pop' phenomena. Press the 'Answer' tab shown.

    https://www.thoracic.org/professionals/clinical-resources/sleep/sleep-fragments/electrode-pop.php

  • Radha_openbciRadha_openbci United Kingdom

    I tend to change boards if one behaves weirdly. I do get noises-same or different in both. I think I agree with you that board isn't the problem. I do place my reference electrode on the left/right mastoid, so that explains the issue. MAybe whenever I notice the issue I should reapply the reference electrode.
    Thanks, it gives some closure although why is it always 1 dual peak in say 5-10 minutes and not single peaks regularly if the reference is bad?

  • wjcroftwjcroft Mount Shasta, CA

    When the reference is having issues, it is usually mirrored in all the channels, not just a single channel. In the graph of yours that I pasted above, this is the case. Although the effect in channel 1 is hard to see; it is still there. The different noise amplitudes are due to the distance of that electrode from the reference.

    I don't understand your statement:

    I do place my reference electrode on the left/right mastoid, so that explains the issue.

    Are you saying that ear lobe would easier, this is likely true. Mastoid reference should be fine, as long as the hair is out of the way and the contact is good. If you look at some of the previous comments on this thread, I mentioned that hair conditioners used after shampoo, can sometimes cause issues with electrode pop. Hair should be shampooed but not conditioned.

    In the past for neurofeedback EEG, also, NuPrep skin abrasion was used. May possibly help in your situation. See the second video on this page:

    https://openbci.com/forum/index.php?p=/discussion/176/placing-electrode-sensors-on-the-scalp

  • Radha_openbciRadha_openbci United Kingdom

    I use NuPrep abrasion before applying Ten20 gel. Yes, I am saying that my peaks could be due to the reference electrode on the mastoid. My reference impedance is generally low. Probably I don't place it on the correct bone every time so could be contact or pressure issues? I can take care of that next time if it happens again to confirm this. I can also confirm to see which side the reference was and which electrode had the peak. But I don't understand the dual peaks?

  • wjcroftwjcroft Mount Shasta, CA

    Try some tests with the reference and Bias placed on the ear lobes. Those are generally easier to deal with.

    But I don't understand the dual peaks?

    Once an electrode pop occurs, it means more pops are likely. The peaks simultaneous across channels are due to the shared reference.

  • Radha_openbciRadha_openbci United Kingdom

    I don't use the Bias pin, what is that for?

  • wjcroftwjcroft Mount Shasta, CA

    Sorry, I meant to say 'Ground' or 'D_G' (driven ground). On Cyton Bias is the same as Ground / driven ground.

Sign In or Register to comment.