Strange high-frequency pattern and systematic amplitude modulation in raw data (Cyton+Daisy)

Hi,

we recently acquired a 16 channel Cyton+Daisy board, using the Bluetooth dongle to connect to a Win10 machine.

When viewing the raw uV values (using MNE python for plotting), I notice a very strange pattern:
image

The raw data values in the .txt file have very high variation sample-to-sample (up to 100uV) and always present this perfectly alternating direction of change between each sample. then there is the strange amplitude modulation... Is this normal?

in the OpenBCI GUI, the data looks fine I think:
image

But this could simply be an effect of the bandpass filtering. Applying the same 1-50 filter within MNE gives a better-looking graph:
image

But you can still see the weird artifacts on P3 and P4 for example, and I'm still convinced the raw data should not look like the first picture.

Any ideas why? I can provide pictures of our Cyton-Daisy wiring but I've gone through the docs multiple times and I think they are correct. Could this be related to the way the boards do an alternating sampling of the 8 Daisy channels and then the 8 Cyton channels? Would this mean we got a bad board?

Here's a screenshot of the raw text file at the corresponding time so you can see the pattern really exists in the numbers
image

Thank you for your help, we are very excited to be able to use our headset with MNE-python to support research in a live setting.

Dan

Comments

  • wjcroftwjcroft Mount Shasta, CA
    Dan, hi.

    The GUI applies both a bandpass and a notch at the mains frequency. Are you doing the same with your MNE? The CSV raw data file does not have any filtering, you must provide that.

    Regards,

    William

  • wjcroftwjcroft Mount Shasta, CA
    edited October 2018
    A Butterworth bandpass filter that goes up to 50 Hz, still leaks some energy through at the band ends. Depending on the filter order. That is why the notch is used.

    See the second image,


    image
Sign In or Register to comment.