Stabilize SD Card Recording

Dear OpenBCI Community,

I am currently running a study with the Cyton+Daisy combo, in which participants wear sensors over the course of a workday. To record data continuously, I am writing the recordings to SD cards. The process is simple:

  1. In the morning, I set up the system using OpenBCI GUI (v. 4.2.0) - i.e. check impedances, etc.
  2. After setup, I restart the amplifier and the software and start a recording with 24h SD card recording
  3. Then, I quit the software (without stopping the recording)
  4. The participants head home afterwards

In some cases the process works well and data is continuously written to the SD card (up to 8 million samples...) that is retrieved in the evening. In other cases (probably 1 of 3 instances), however, the SD card recording fails and only 2000 to 6000 samples are recorded. I tested the SD cards before and determined that they are generally able to record at least 80k samples - and as I said, it mostly works well.

However, I wonder where the volatility is coming from, i.e. why a third of the recordings stop short. Would someone happen to have experiences with this issue? Or perhaps an idea on how I might stabilize the recordings? Could it be "just" an issue of power interruptions (e.g. some participants ride their bikes home while wearing the system - which might cause brief power interruptions)?

Thank you in advance for your feedback!

Best regards
Michael

Comments

  • wjcroftwjcroft Mount Shasta, CA

    Michael, hi.

    Is there any correlation to the type or capacity of SD card being used? There is some speculation that larger capacity SD cards may have less reliability.

    https://forums.tomshardware.com/threads/are-smaller-64gb-micro-sd-cards-more-reliable-than-larger-128gb.2773399/
    https://www.google.com/search?q=are+smaller+sd+cards+more+reliable
    https://docs.openbci.com/docs/02Cyton/CytonSDCard

    William

  • MKnierimMKnierim Germany

    Hi William,

    thanks for the quick reply and ideas! So far, I don't see a relationship between SD card capacity and the recording loss. I am using four boards for the experiment and mostly 32gb SD cards (same model - as recommended in the documentation). In one instance, I did use a 64gb card, but that worked fine.
    I also see right now that two boards are working fairly reliably. In one board, I do see that there is a slight issue with temporary power connection loss. That is, the slightest touching of the power lever causes the power light to turn off (or flicker). So perhaps this effect is responsible? Would you happen to have an idea how I could handle this?

    Best
    Michael

  • wjcroftwjcroft Mount Shasta, CA

    There are spray contact cleaners that can restore proper conduction in switches. These also remove oxidation and contact degradation. Check with your local electronics tech person, they may have something similar. This is the one we use here in the US, it works excellently. You would use the tube attachment to target the opening in the switch where the 'lever' pokes out.

    https://www.amazon.com/gp/product/B00006LVEU/

    So perhaps this effect is responsible?

    Yes, any momentary power loss will cancel your recording.

    William

  • wjcroftwjcroft Mount Shasta, CA

    Another point to note, is that the Cyton switch has three positions: PC, off, and BLE. As far as the firmware goes, PC and BLE are equivalent, since Cyton firmware has no 'BLE' feature. It's possible that the switch internal mechanical contact is more worn on the side you usually use (probably PC). Thus by using the BLE side instead, the contact oxidation there may be less, and more reliable.

    It's still a good idea to try a decent de-oxidizing contact cleaner as well. (Such as the De-Ox-it link mentioned. Some cleaners just contain lubricants, and they are unlikely to restore operation.

  • retiututretiutut Louisiana, USA

    I've added a link to this Forum thread to an open issue in the GUI repository. https://github.com/OpenBCI/OpenBCI_GUI/issues/893

Sign In or Register to comment.