Recording crashes with SD card and other issues

Hi there,
I tried to record several hours but I ran into some problems:
Setup:

  • Cyton+Daisy Board with 16 channel OpenBCI electrode cap
  • SD card 64GB
  • USB Dongle on Windows PC
  • Configuration in GUI: no file size limit for recording on local PC, 12h limit for recording on SD card
    Everything was set up according to the OpenBCI Tutorials/Getting Started Guides

I tried a pilot with some 5 minute recordings and ran some issues. Then I tried to record 4h and recording data crashed after a few minutes:

  1. The incoming signal was very choppy. I fixed the FTDI Buffering according to "Fixing FTDI Buffering on Windows" section in OpenBCI docs. The signal was much cleaner, but after a few seconds, there occured several spikes in amplitude. I can't find this issue in the recording but it is very annoying when you inspect your data live and there are these spikes all the time.

  1. Often, there occurs a notch at 25Hz (see image). Sometimes, the notch occurs only when recording with 16 instead of 8 channels, sometimes it also occurs with 8 channels, sometimes it doesn't occur. The notch is also in the recorded data. However, the 25Hz notch is only in the data recorded via the dongle and not in the data recorded on the SD card. I found this article: http://openbci.com/forum/index.php?p=/discussion/889/unexpected-noise-with-daisy . The problem there was that some leads were "on air" and not connected properly. However, this is not helpful in this case as the notch is only in the data recorded via the dongle. How can I get rid of this notch and why does it exist?

  1. Sometimes there are electrodes that are zero all the time. I am sure the electrodes are connected properly. What could be the reason for this? After a restart of the board, the electrodes usually work again but sometimes they also crash during recording.
  2. When recording longer than a few minutes, the recording crashes on the SD card. I recorded 4h and after evaluating the SD card, I noticed that in there are only 13930 rows filled with sensor data (2 minutes). What could be the reason for this? I formatted the SD card with the OpenBCI guide. When using the GUI for converting the file, the GUI crashes and is not responding after I selected the file.
  3. Then I tried to inspect the PC recording. The GUI said that the recording is only 741s long (expected 4h). After about 370s the stream starts accumulating values or squeezing the time series data. In every second, the stream shows values from more than 10s. On the picture you can see that in about 5 minutes, more than 1h of recording is accumulated. Then the recording stops and the other 3h didn't became recorded. What could be the reason for this? How can I fix it?


I think number 4 is the most important point for me as it prevents the most from doing a good recording. I am very pleased for any ideas and help!

Best,
Georg

Comments

  • wjcroftwjcroft Mount Shasta, CA
    edited May 2020

    Georg, hi.

    re: "...The signal was much cleaner, but after a few seconds, there occurred several spikes in amplitude..."

    If you are seeing occasional radio packet loss during a session, this can cause glitches in the time series. Especially if you are using an older version of the OpenBCI_GUI and Hub. Newer GUIs use 'interpolation' to avoid such behavior. But the best solution is to improve the radio link reliability between the dongle and mainboard. Such as (1) changing the radio channel. (2) using a usb extension cable to get dongle closer to mainboard. (3) getting your equipment into a quieter microwave environment, away from other wifi, Bluetooth, cell phone equipment. (4) also positioning OpenBCI away from other sources of mains noise such as extension cords, power supplies, conduits (in floors or walls), etc.

    re: 25 Hz peak in FFT.

    If your mains frequency is 50 Hz, the 25 Hz would then be considered a 'sub-harmonic'. And just another manifestation of mains noise. You MUST have good scalp connection, decent impedance on all your channels that are in use. Check impedance by clicking the little 'ohm' symbol next to the channel number. It should be at most a couple megohms, any higher than that, and the mains noise effects will be difficult to suppress.

    re: bad recordings

    I would only suggest trying another SD card, perhaps smaller in capacity, and as highly rated as you can.

    Regards, William

  • Hi William,
    Thanks for you help. I will try this.
    The impedance is about 50kOhm. Is this correct?
    What does it mean if some electrodes record 0.0 (and 0.0Ohm) all the time? How can I fix it?
    I am using a SanDisk Ultra as suggested. Any other ideas why recording is unstable and fails?
    Best regards,
    Georg

  • wjcroftwjcroft Mount Shasta, CA

    50K ohms is great. If some channels show zero microvolts and zero ohms, that means that the channel is apparently 'shorted out', that is, the wire for the channel has some kind of direct connection to the reference lead. This could only happen if there is some error in wiring, or possibly a situation involving an "electrolyte bridge" with gel, saline or other conductive material leaking on the scalp.

    I would try another SD card and reformat it. It's possible your particular SanDisk card is damaged or defective somehow. Though in general SanDisk is a good brand. Perhaps even try another brand. There are sometimes counterfeit cards on the market. Don't go for a huge capacity, just only how much you need. Be sure the card handles at least 10 MB/s.

    https://kb.sandisk.com/app/answers/detail/a_id/1996/~/difference-between-speed-class,-uhs-speed-class,-speed-ratings-(performance)

  • I tried with a 16GB SD card now that fullfills the requirements. But unfortunately, I have to confirm that the problems are still there.

  • wjcroftwjcroft Mount Shasta, CA

    Does the recording work with 8 channels, but fails with 16?

  • wjcroftwjcroft Mount Shasta, CA

    When was this purchased? Recently? Has it ever worked correctly? Do you have the latest firmware, and are using the latest GUI? Have you tried running on another laptop?

  • I have the latest GUI and the OpenBCI was purchasted very recently but it didn't ever work correctly . I tried with 8 instead of 16 electrodes and a 2000s long recording seems to was successful.

  • wjcroftwjcroft Mount Shasta, CA

    Have you tried using it with another computer? Have you resolved the issue with the zero microvolts zero impedance channels? The fact that these reset to normal after power cycling the board, is suspicious of broken hardware. Is that still happening, even with 8 channels?

  • wjcroftwjcroft Mount Shasta, CA

    If the tests using only the 8 channels on the mainboard (Daisy unplugged), are all successful -- then it is possible that the 16 channel recording and other glitches you are seeing are due to a flaky Daisy board. If that is the case, send an email to contact at openbci.com and explain your situation with the suspected Daisy board.

  • I tried 4 recordings today with 5-60 minutes recording time. 2 were successful recorded, from one recording, the last 10 minutes are missing and the 4th recording failed. So it's probably not the Daisy board. Is there any way to get immedately notified if recording fails?
    With the 8 channels recording, there didn't occur any problems with 0.0uV today.
    Tomorrow, I am going to try it with another computer.

  • wjcroftwjcroft Mount Shasta, CA

    So your 4 recordings were all 8 channel? And two failed. It could be that the zero microvolts zero ohm issue is aggravated by having the Daisy plugged in, but your recording issue is on the mainboard alone. If the trial on the other computer shows the same behavior, that would pretty certainly implicate the mainboard hardware. Especially the SDcard interface.

  • Hi, Yes the 4 recordings were 8 channel.
    I tried today on a few other computers with Windows 10 and new Hardware, but following problem occured: After inserting the dongle, Windows recognizes the device and the dongle's blue LED turns on. But after 10s, the dongle turns off and I can't connect to the openBCI (but it still works on my laptop). Is there anything, I have to configure on the usb ports of the computers?

    Also, I did the observation that usually channel 8 and 16 are at 0.0uV. When I start a session and in the session turn off and on the device, it usually works again. This would be fine for me as a temporary workaround but the problem is that no data is saved on the SD card after restarting the device in a running session although data is received on the laptop. (Is this the intended behavior?) Also, when start and stop the data stream in a running session multiple times, only the first data stream is saved, the others become discarded. (Is this the intended behavior)?

  • wjcroftwjcroft Mount Shasta, CA

    As far as installing the drivers for the dongle: on newer Windows 10 machines it should be automatic. But on older machines you may need to download the VCP driver from FTDI:

    https://www.ftdichip.com/Drivers/VCP.htm

    re: your SDcard issues. It appears from your many tests that the SDcard hardware on your board is defective. Especially since this is a brand new board, email contact at openbci.com and ask them how to proceed, and if any more testing is recommended. Since you've already tried other SanDisk cards, and with/without Daisy, and the 0uV issue -- your mainboard seems very suspect.

  • Thanks for your help, I'll do that.

  • CidCid Vancouver

    Has anyone had issues with the sd card only saving files half of the time? For me it seems like only every second sd card recording is saved on my sd card.

  • I had the same issue with saving EEG data onto SD card of Cyton board. It really saves the data about every second recording (sometimes it doesn't save). Does anyone knows how to fix that? Thanks

  • Hi there,

    I experienced the same issues! Although I had not the time to check the real reason for this, I developed a few techniques that seemed to raise the chance of a successful recording:

    • Don't delete files on the SD card. Often, it did not record after I erased all my files.
    • Before a recording, restart the GUI and do one test (a normal recording) to check if it it records properly. If so, continue with the experiment.
    • Do not record longer than 20 or 25 minutes and use "record 60 mins max on SD-card" option.

    Although these steps look rather random, they helped me to increase the number of successful recordings significantly, but a few ones failed nevertheless.

    Of course, it would be great if anyone of the OpenBCI developers could look into this issue.

    Best,
    Georg

  • retiututretiutut Louisiana, USA
    edited October 2020

    We can try to look into this, and I will share with the team. Thank you for the additional information @GeorgLange!

    Making an issue on the GUI for now until it can be determined if this is a firmware or software issue and linking to this thread.

  • retiututretiutut Louisiana, USA

    Issue has been logged on Github for this thread: https://github.com/OpenBCI/OpenBCI_GUI/issues/893

Sign In or Register to comment.