OpenBCI GUI can't connect to Cyton board

Comments

• QWERTY12345 Canada

  January 2020

  I tried updating the firmware by following the instructions here but I got this error :

  Arduino: 1.8.10 (Windows 10), Board: "OpenBCI 32"
  Sketch uses 84904 bytes (69%) of program storage space. Maximum is 122880 bytes.
  Global variables use 10564 bytes (32%) of dynamic memory, leaving 22204 bytes for local variables. Maximum is 32768 bytes.
  An error occurred while uploading the sketch
  \\.\COM5: Cannot open
      Copyright: (C) 2011-2015 Serge Vakulenko
  
                     2016-2017 Majenko Technologies
  No target found.
  Programmer for Microchip PIC32 microcontrollers, Version 2.1.46
  
  This report would have more information with
  "Show verbose output during compilation"
  option enabled in File -> Preferences.
  
• wjcroft Mount Shasta, CA

  January 2020 edited January 2020

  Hello Q,

  Suggest updating the OpenGL graphics driver software on your laptop. This is especially important for older hardware, what is the manufacturer, model, date of your laptop?

  https://www.google.com/search?as_q=opengl&as_sitesearch=openbci.com

  OpenGL libraries are used by the Processing Language, in which the GUI is written.

  Regards, William
• wjcroft Mount Shasta, CA

  January 2020

  Actually I take back that comment about the OpenGL possibly being the root of the error. None of the lines referencing OpenGL in the backtrace are showing an error status, only that that an opengl routine was called.

  Instead it looks like the GUI is hitting some snag as it is trying to do a network write (of commands) to the Hub. Then the connection is broken and the null pointer results. But earlier it does say that the Hub was started ok.

  Are you sure that you followed the instruction to set the Hub program exe as "Run as Administrator"?

  It looks like you are running from Processing. Have you tried the standalone?

  William
• QWERTY12345 Canada

  January 2020 edited January 2020

  Thanks for the reply!
  I was pretty sure that I have the standalone version, I'll try installing it again
• erisian

  January 2020

  I am having the same issue as the OP, I think.
  Also have same versions of GUI and FTDI Drivers... Windows 10 a slight newer at 10.183623
  I can connect to the Cyton by manually selecting the com port (happens to be 8, or 19, I have two dongles).
  Autoscan will find the Cyton on channel 2.
  Associated Log looks like this...

  Radios_Config: system_status
  Radios_Config: Success: System is Up
  [SUCCESS]: Successfully connected to Cyton using COM8
  Radios_Config: system_status
  Radios_Config: Success: System is Up
  [SUCCESS]: Successfully connected to Cyton using COM8
  Radios_Config: get_channel
  Radios_Config: Success: Host and Device on Channel number: 2
  

  But... as soon as I try to start a session... it fails... on "init Timeout".

  I note the same complaint in the log as the OP... that the config file cannot be found.

  The file "C:\Users\tim\Documents\OpenBCI_GUI\Settings\CytonUserSettings.json" is missing or inaccessible, make sure the URL is valid or that the file has been added to your sketch and is readable.
  InitSettings: C:\Users\tim\Documents\OpenBCI_GUI\Settings\CytonUserSettings.json not found or other error.
  [DEFAULT]: The new data source is LIVE (from Cyton  ) and NCHAN = [8]. User Settings Error: Default Settings Loaded.
  

  However, the file exists...
  It says use com port 3.
  I edited it by hand to say COM 8... but it gets overwritten again every time I try to start a session... back to COM 3.
  I think something is going wrong with the session config update, it's failing to default settings and not seeing the Cyton on that COM Port.

  I also tried this on a windows 7 VM... and got the same results exactly... except the com port numbers were different.

  Any suggestions, appreciated.
• erisian

  January 2020

  Another observation...
  If I click on settings, I have only one option... "Clear All"
  (I would expect more, given the hover-help: "Save and Load GUI Settings... etc...")
  For what it's worth, I opened ports on private and public newtorks when prompted...
• retiutut Louisiana, USA

  January 2020 edited January 2020

  @erisian Those settings aren't relevant until you start a session.

  .. that the config file cannot be found.

  When no config file is found, it will say so in the log.
• retiutut Louisiana, USA

  January 2020 edited January 2020

  If you run from Processing, please make sure to use the version of the OpenBCI Hub associated with that version.

  GUI 4.1.7 uses Hub 2.1.0.
  https://www.GitHub.com/openbci/openbci_hub/releases/latest

  The next release of the GUI will not use the Hub.
• erisian

  January 2020

  Thanks for the feedback, retuitut.
  I get that the menu items might not be relevant until a session is established.
  The failure here is at the point of establishing the session, so, I guess they wont be visible until that happens.
  I am not using Processing. Everything I have been doing is with the standalone application (so hub version is aligned).

  So... what else can I offer troubleshooting...?

  • Clearly, given that the \documents settings file is being overwritten at the time of attempting to establish the session, the folder is writable (confirmed timing by having it open in Notepad++ which is non-locking but reports file modification by dialogue pop-up).
  • Perhaps worth noting... Same results when run as Admin or not.

  I have perfect reproducibility with:

  • each of two different dongles (resolving to distinct COM Ports, but otherwise behaving the same).
  • each of two different computers (my laptop running Win 10 as stated earlier and a Win 7 VM with USB passed through)
  • multiple versions... version 3.4.0 and version 5.0.0 behaved same, for me, as the current release.
    (one advantage of standalone application is the ease of testing this... public and private ports opened on each first execution)

  If I power down the ultra-cortex and try to autoscan the leds flash for 15 seconds or so and the link fails, as would be expected.
  If I power up the ultra-cortex and try to autoscan it always fails on first attempt with a fresh start but always works on second attempt, resolving a link on Channel 2.

  Its when I try to initiate a session that the wheels come off...

  Cheers.

  Dear OP (Qwerty) - I hope we have the same problem!!! Sorry for hijacking your thread if not!!!
• erisian

  January 2020

  @retiutut said:
  @erisian Those settings aren't relevant until you start a session.

  .. that the config file cannot be found.

  When no config file is found, it will say so in the log.

  It does say something to this effect... it also says it in the OPs log... which was one of the reasons I thought our issues were same/related:

  The file "C:\Users\tim\Documents\OpenBCI_GUI\Settings\CytonUserSettings.json" is missing or inaccessible, make sure the URL is valid or that the file has been added to your sketch and is readable.
  

  Cheers.
• retiutut Louisiana, USA

  January 2020 edited January 2020

  Ok. The only way they would be the same is if this appears:

  OpenBCI_GUI::Setup: Setup is complete!
  Instantiating hub object...
  Hub: CheckHubInit: The GUI is connected to the Hub!
  Client SocketException: Connection reset
  

  This SocketException is happening outside of the GUI code in a dependency.
  To fix this, we have been working since last year to make a GUI without the Hub. This is actively in progress to resolve issues like the above.

  In conclusion, I have a hunch that the next GUI release that uses a new Java library, instead of the Hub, will fix any type of Socket issues, as it would no longer be used.
• retiutut Louisiana, USA

  January 2020

  @erisian Here is a little guidance to start troubleshooting your issue:

  If I power down the ultra-cortex and try to autoscan the leds flash for 15 seconds or so and the link fails, as would be expected.
  If I power up the ultra-cortex and try to autoscan it always fails on first attempt with a fresh start but always works on second attempt, resolving a link on Channel 2.

  Yes, I just added a message to say "Please press autoscan again" when I found it needs to be pressed twice.

  Its when I try to initiate a session that the wheels come off..

  Go ahead and make a new post in the Forums or OpenBCI GUI Issue on GitHub and include a full console log file and maybe some screenshots.
• QWERTY12345 Canada

  January 2020

  Hi,
  Sorry for the inactivity, I almost forgot about this post. We weren't able to make it work but we did manage to figure out a workaround with pylsl: https://github.com/chkothe/pylsl
• retiutut Louisiana, USA

  January 2020

  @QWERTY12345 So you are using pyOpenBCI and pyLSL?
• QWERTY12345 Canada

  January 2020

  A combination of these two, my bad
  https://github.com/labstreaminglayer/liblsl-Python/tree/master/pylsl/examples
  https://github.com/OpenBCI/OpenBCI_LSL
• retiutut Louisiana, USA

  January 2020

  Can you share a simple example here? Maybe with just Time Series data format.
• QWERTY12345 Canada

  January 2020 edited January 2020

  Sure:
  1. Download the repository https://github.com/OpenBCI/OpenBCI_LSL and the .py files from https://github.com/labstreaminglayer/liblsl-Python/tree/master/pylsl/examples in the same folder
  2. Run the command " python openbci_lsl.py --stream " or " python openbci_lsl.py [PORT] --stream " if you get an error
  3. Run ReceiveData.py
  Here's an example of code that I wrote, using existing code from ReceiveData (it's not very optimized or anything, I was just experimenting with the library since it was my first time using it) :

  from pylsl import StreamInlet, resolve_stream
  import numpy as np
  import time
  def saveData():
      print("looking for an EEG stream...")
      streams = resolve_stream('type', 'EEG')
      # create a new inlet to read from the stream
      listTimestamp = []
      c1 = []
      c2 = []
      c3 = []
      c4 = []
      c5 = []
      c6 = []
      c7 = []
      c8 = []
      lenght_data = 5000
      nb_data = 0
      inlet = StreamInlet(streams[0])
      n_sample = 0
  
      while True:
          begin = int(time.clock() * 1000)
          # get a new sample (you can also omit the timestamp part if you're not
          # interested in it)
          sample, timestamp = inlet.pull_sample()
          # append the sample of timestamp and each channel to the list
          listTimestamp.append(timestamp)
          c1.append(sample[0])
          c2.append(sample[1])
          c3.append(sample[2])
          c4.append(sample[3])
          c5.append(sample[4])
          c6.append(sample[5])
          c7.append(sample[6])
          c8.append(sample[7])
  
          nb_data += 1
  
          # save when there is 5000 timestamps
          if nb_data > lenght_data:
              # reset the number of data
              nb_data = 0
              # save in CSV
              all_data = [listTimestamp, c1, c2, c3, c4, c5, c6, c7, c8]
              np.savetxt("sample" + str(n_sample) + ".csv", np.asarray(all_data), delimiter=",")
              # reset
              listTimestamp = []
              c1 = []
              c2 = []
              c3 = []
              c4 = []
              c5 = []
              c6 = []
              c7 = []
              c8 = []
              n_sample += 1
          end = int(time.clock()*1000)
          print(end - begin)
  
  saveData()
  
• erisian

  January 2020

  Nice share, Qwerty