OpenBCI "Clipping" Issue

naznaz Italia
in Cyton
Hello,

I’m practising with the Cyton board (firmware version 3.1.1) to assess its signal quality (for now without daisy).

In principle, I’m interested in using only the WiFi Shield (firmware version 2.0.5), but I experience the following issue also with Bluetooth and both with and without SD card.

As it is shown in the picture below (and even more clearly in the videos linked later), sometimes the signal seems to clip and remains stuck until I shake the board or move the cables. In addition, this behaviour changes from acquisition to acquisition, and in some rare cases, it self-recovers without the need of shaking the board. 
From my tests (acquisitions of a single ECG bipolar channel as presented in the Cyton tutorial, see videos for configuration), it seems to happen when I move rapidly towards the board (in this case it happens all the times) or when I move the limbs (but not always, for long intervals it doesn’t happen even if I move a lot even the cables. Of course, there’s the expected introduced noise, but no clipping). 

Note that I’m testing the board with movements because they will be part of my experiments. Motion artefacts are expected… I will deal with them somehow, but I cannot deal with such long clipping intervals (seconds) especially because most of the times the board doesn’t self-recover.




What I really don’t understand is where this behaviour originates and why sometimes the board works perfectly while immediately after, with the same movements, it does not. My guess is that it is something happening in the ADS1299, but I really cannot spot the source (I don’t have strong electronics experience but I read the datasheet various times and I have no clue), especially considering that to solve the issue I have to move the board (and sometimes it doesn’t even work and I have to do it many times).

Can you give me some hints? Some possible explanations? Solutions?

NOTE1: I experience this issue with two different Cyton boards with the same firmware version (I don’t think it’s a firmware problem).
NOTE2: As the picture shows, it is not actually clipping, because there is an interval (slightly before the 300000th sample) with values that exceed the value experienced during the issue (and also the “clipping” value is not perfectly constant as can be seen from the linked file, but still seems to be clipping from the below image of the induced recovering beginning). I’m really confused. 






NOTE3: If I start an impedance test on the channel when the board is stuck (and is not self-recovering) it gives 0 ohms as if it senses it as RAILED.
NOTE4: In the wires configuration of the acquisition in the videos the BIAS pin is not used (and removed along with SRB2 in the hardware configuration of the OpenBCI_GUI), however, I also tried with it included and placed somewhere in the arm and I experienced the very same behaviour.

VIDEO2: https://youtu.be/0_1MC4NCrAo <— here the issue is very clear
VIDEO3: https://youtu.be/V74JwBSiQOw <— more stable behaviour


Some notes and reasoning on the videos:
  1. In this specific acquisition, it always self-recovered from the clipping (in other acquisitions it did not). Actually, in this acquisition, it never seems to actually clip, but rather to have a large swing in amplitude. (this might be the difference from self-recover w.r.t. no self-recover, need to investigate further)
  2. As can be seen at the beginning of VIDEO1 (until minute 00:50), moving the legs (and consequently the wires), feet and hands does not produce the clipping, even manually moving the wires does not. This is the intended behaviour. The issue happens at minute 00:50 when I rapidly move (but not so much in the end) closer to the desk a couple of times. Then again no issue even moving until minute 01:54 when a couple of usual leg movements generate the issue. Then same movements no issue until the end of the video.
  3. The first interval of VIDEO2 presents the issue due to slight leg movements again (until minute 01:22). Then no issue during manual movement of the wires (this puzzles me).
  4. VIDEO3 has a more stable behaviour (just one instance of the issue) even if it was acquired immediately after VIDEO2. 

Thank you in advance for your support. I'm planning new acquisitions to try to constrain the problem trying to get more insights, so if you have any suggestion please feel free to reply.

Comments

  • wjcroftwjcroft Mount Shasta, CA
    Naz, hi.

    I just looked at your first two videos. While I do see some large variations (you call 'clipping'), these look like normal movement artifacts to me. What electrode are you using, I think it is on your arm? You might find better quality using a foam adhesive Ag-AgCl electrode for EMG, if you are using a cup electrode and paste. Or if you already use a foam electrode, try another brand.

    What happens when you follow exactly the tutorial instructions to measure ECG, EMG, EEG?


    Regards,

    William

  • naznaz Italia
    edited July 2018
    Hello,

    First of all, thank you for the fast reply.

    Then, yes, in the acquisition session captured in the video it is not actually “clipping”, but if you take a look at the picture that I linked (which corresponds to another acquisition session) it is actually clipping, so the issue is there. If I collect data during my experiments and it clips and doesn’t self-recover I have to throw everything away (if I’m not monitoring the signals).

    I additionally linked the videos trying to provide as much information as possible, especially because the clipping in the other sessions happened exactly with the same movements shown in the video (which I replicated on purpose to show the issue). Unfortunately, in the other acquisition sessions I wasn’t filming, but tomorrow I will acquire new data and I will take new videos.

    Getting back to your reply, I agree that they are motion artifacts, but they are strange. The point is that doing always the same movements they appear only sometimes. I don’t understand why, as you can see in the videos, I have long intervals with good data (with some normal motion artifacts) and sometimes with the very same movements the signals processed by the Cyton swing so much (in some cases resulting in clipping as in the other sessions).

    I’m doing these tests because I need to stress the OpenBCI with this type of movements since I will be recording while driving. Unfortunately, in this condition, I cannot use it, but it’s strange because for there are periods in which it works very well even with movements.

    Tomorrow I will provide more information to better clarify the behaviour, possibly recording also with another acquisition system.

    To reply to your questions: I’ve been following exactly the tutorial you pointed out to measure ECG (indeed the sensors are placed on left and right arms) using two different types of adhesive AgCl electrodes (the ones sold in the OpenBCI shop and some others from Neuroelectrics), and also tried with the OpenBCI gold cups with paste. The behaviour is the same with all of them.
    I haven’t tried EEG yet because I wanted to stick with the simplest possible configuration (i.e., with known and expected signal shape), but I will try to acquire EMG and EEG as well, even if I don’t think it is relevant for explaining this specific behavior (apart maybe considering a setting with shorter wires).

    I hope we can continue the discussion because I’m not convinced with just the motion artifact option. :-)
    Just brainstorming, do you have any possible alternative hardware/software related idea?


    Best Regards,
    Andrea
  • wjcroftwjcroft Mount Shasta, CA
    Andrea,

    In the tutorial, Bias (Ground) is connected, even though the ECG setup is + - differential. You always need a Bias connection in any measurement. Where is your Bias?

    I suggested following ALL the steps of the tutorial, because it covers many bases.

    William
  • naznaz Italia
    Yes, I know since I read many of your answers in the forum where you explain why we always need the BIAS, and indeed, as I have written in NOTE4 in the original post I also tried with the BIAS as specified in the tutorial, but the result was the same (it increases signal quality, but it does not solve this particular issue).

    In any case tomorrow I will include it again.


    Andrea
  • wjcroftwjcroft Mount Shasta, CA
    Also check your cables by wiggling the connector-wire junction at each end. You can also do this with an ohmmeter to check for zero ohms. Try swapping around your cables, for example to use a pair you have not in the original trials.
  • naznaz Italia
    Ok, I will try all these hints. 

    Thanks
Sign In or Register to comment.

© Vanilla Keystone Theme 2026