Spontaneous Spikes in OpenBCI GUI

eegstud24eegstud24 Williamsburg, VA

Hello everyone!

I need a bit of help. I have the Ultracortex headset with the Cyton. Recently, when I've been trying to use the GUI, there are a bunch of transient spikes in the Time Series widget. I'm running the GUI on Windows 10, though this same issue persists on my Mac. Overall, the BCI works, as verified with eye blinks and alpha wave tests, but these spikes interfere with classification for experiments like P300, for example. Some sessions don't have the spikes at all, but others do.

I've tried a couple of troubleshooting steps, including:
1) moving to a different spot in the room with presumably less electromagnetic interference
2) replacing electrodes
3) taking the usb out, powering off Cyton, putting usb back in, powering on Cyton

If anyone can take a look at the attached YouTube link and help me troubleshoot, that'd be great. Skip to 0:15 to see the spikes in channels 7 and 8. I've slowed down the video at that point. I've also attached a photo of me wearing the headset.https://youtube.com/watch?v=lOgM_Cb_7h8

«1

Comments

  • wjcroftwjcroft Mount Shasta, CA

    Hello WYE,

    Mentioning Richard @retiutut , in case he might have some insights. If this was a radio packet loss situation, one would expect to see glitches on ALL the channels, yet channels 1 and 2 are much less impacted. Though they do have a tiny bit of spiking showing through occasionally that lines up with the large spikes in the other channels.

    Have you checked your impedances with the 'ohm' symbol to the right of each channel? Are you using any kind of hair products or conditioners? Sometimes those can leave a film on the scalp that interferes with EEG. Is this a new Ultracortex / Cyton? Has it ever worked without these spikes? Do the spikes show with other subjects. Etc.

    Regards, William

  • retiututretiutut Louisiana, USA
    edited May 2020

    This looks like packet loss. I recommend moving the Dongle away from the computer with a USB extension cable or USB hub.
    The Cyton battery may also be getting low. A Cyton with batteries about to fail will show huge packet loss.

    Thank you for providing a picture of the setup and a video.

  • eegstud24eegstud24 Williamsburg, VA

    Hello Richard and William,

    It was indeed packet loss. I ran the GUI with Processing and right when the spikes occurred the following lines in the log were printed:
    Hub: apparent sampleIndex jump from Serial data: 249 to 251. Keeping packet. (1)
    numPacketsDropped = 1
    Interpolating dropped packets...
    Hub: apparent sampleIndex jump from Serial data: 58 to 60. Keeping packet. (2)
    numPacketsDropped = 1
    Interpolating dropped packets...
    Hub: apparent sampleIndex jump from Serial data: 194 to 196. Keeping packet. (3)
    numPacketsDropped = 1
    Interpolating dropped packets...

    I'll try the fixes Richard mentioned; I think my battery might in fact be getting really low. Thank you so much for the help you two.

    Best regards,
    Winson

  • wjcroftwjcroft Mount Shasta, CA

    Are there messages printed or counts available in the GUI, to confirm the packet loss? Why would channels 1 and 2 not show the same spikes? I thought the GUI now corrects for single packet loss? So it would have to be multiple packets being dropped?

  • wjcroftwjcroft Mount Shasta, CA

    I mentioned this before to Richard, but some type of indicator on the normal GUI screen (not an invisible log file), would help users to better track when this is happening. Especially if the user's batteries are running low. I know with Pocket-Neurobics gear that BioEra and Bioexplorer main screens can both report both packet loss and low battery. OpenBCI does not have battery level capability, but could indicate packet lossage.

  • eegstud24eegstud24 Williamsburg, VA

    Hi William,
    I don't see any messages in the GUI alerting me to packet loss. The video I recorded doesn't show anything either. As for why only certain channels are affected, I'm not sure either. Sometimes all channels do see a spike, but 1 and 2 are definitely less vulnerable.

  • wjcroftwjcroft Mount Shasta, CA

    Let us know if fresh batteries solve your issue. Or the usb extension cable, it will allow you to place the dongle as close as possible to the Cyton mainboard.

    You are correct, the GUI does not currently show this on the main screen. Only the log, which is usually not shown. On some other neurofeedback systems I use, the packet lossage is highlighted on the main screen with some type of indicator (count or flashing dot.) These systems also show battery level, but OpenBCI hardware does not make that available.

  • eegstud24eegstud24 Williamsburg, VA
    edited May 2020

    Unfortunately, I still get some spikes here and there. It is truly strange how channels 1 and 2 aren't affected.... all the electrodes (except the forehead ones, which correspond to channels 1 and 2) are the dry comb FRI electrodes, and upon inspection none of them show any significant wear. I've recharged the battery, and I put some distance between the computer and the dongle by using an extension cord. Please see the attached video. I still think it's packet loss because of the log, but I don't know what the remedy would be.

    https://drive.google.com/file/d/1wg3yb1GM9ayuTBxbmOUk1WpNh9W1ePcr/view?usp=sharing

  • wjcroftwjcroft Mount Shasta, CA

    The idea with the usb extension cord, is PRIMARILY to get the dongle as close as possible to the Cyton mainboard. Not so much to "put some distance between the computer and the dongle." Could you try pinning the dongle to your shirt for example? Have you tried moving to other areas that might have different EMF (electromagnetic field) interference, or mains noise interference?

    Has this Cyton EVER worked well without the spiking? When did you purchase it?

  • wjcroftwjcroft Mount Shasta, CA

    Impedance for all your channels looks reasonable. Channel 5 is on the high side. But no impedance issues can result in packet loss.

  • eegstud24eegstud24 Williamsburg, VA
    edited May 2020

    I'm a bit confused. Richard recommends "moving the Dongle away from the computer with a USB extension cable or USB hub"? Nevertheless, I've tried situating the dongle close to the Cyton, and unfortunately the result is the same. Yes, I've tried other areas with less EMF, and verified through FFT that the 60 Hz line noise in those areas is considerably smaller. In the beginning, I think the Cyton also had these transient spikes, but they were so transient I kind of just ignored them, as I was able to get alpha waves and eye blinks. The purchase was made back in January.

  • wjcroftwjcroft Mount Shasta, CA

    The closer the dongle is to the mainboard, the higher the signal strength of the Bluetooth radio packets between the two RFduino radios. So the distance between the two RFduinos is generally the most critical factor in reducing packet loss. Separating the dongle from the laptop will also reduce noise, if the noise is coming from other EMF emissions from the laptop or power supplies nearby.

    One other suggestion: is the packet loss situation any different, if you unplug the laptop AC wall power supply from the wall? So the laptop is entirely running on battery? If there is an odd grounding situation with your laptop / power supply, it can increase noise sensitivity.

    You said this behaves exactly the same as with your Mac, as with Windows?

    It's possible that one or the other RFduino's (mainboard or dongle) has a marginal radio transmitter / receiver. Especially if this packet loss was present on the date when it arrived new.

  • eegstud24eegstud24 Williamsburg, VA

    Yes, on a Windows desktop the noise is still present. My laptop is running entirely on battery. Hmm.... would I have to replace the Cyton or dongle in that case?

  • wjcroftwjcroft Mount Shasta, CA

    This is looking more like you have received a defective Cyton in January. Especially since the packet loss persists on multiple computers and locations and regardless of dongle-Cyton proximity. I suggest you email contact at openbci.com and ask how to proceed. A new dongle would be easy to send. If that is not sufficient the mainboard RFduino may be the culprit.

    It sounds like your firmware is functioning just fine. So a firmware reload would not be of much use. Though the behavior with channels 1 and 2 (mostly not spiking) does seem odd. Maybe Richard @retiutut might hypothesize why that would be happening. Surprisingly the single packet losses, do not seem to be interpolated as expected. Both of these oddities might be a sign something weird is going on.

    Regards, William

  • wjcroftwjcroft Mount Shasta, CA

    One thing I forgot to suggest is to try switching your Cyton radio channel. If you have other Bluetooth radio gear in the vicinity, this can help. Do you have some Bluetooth or wifi gear operating close by? Both those services use the same radio frequencies as the RFduino dongles.

    https://docs.openbci.com/docs/01GettingStarted/01-Boards/CytonGS#5-optional-settings

    Click the arrow in section 5.

  • eegstud24eegstud24 Williamsburg, VA

    Unfortunately, switching the channel yielded the same results. I'll go ahead and email OpenBCI. Thanks for all your help, William!

  • retiututretiutut Louisiana, USA
    edited May 2020

    @eegstud24 Hold on, can you please try using 4 AA batteries and the battery pack that came with the Cyton? It does not need to be connected to a user to investigate packet loss. Let me know if the packet loss is any better using 4 AA batteries.

    I can see that the packet loss is really bad here in the second video. Thank you for showing the console log.

    I'm a bit confused. Richard recommends "moving the Dongle away from the computer with a USB extension cable or USB hub"?

    I use something similar to this: https://www.amazon.com/AmazonBasics-Extension-Cable-Male-Female/dp/B00NH13UFQ/

  • wjcroftwjcroft Mount Shasta, CA

    To underline what Richard is saying about rechargeable batteries -- if you have mixed batteries of different charge history, then older batteries in the pack can distort and warp the performance of the overall pack.

    The usb extension cables are pretty common and all have a male usb Type A on one end, and female Type A on the other end. This would allow you to place the dongle within inches of the Cyton mainboard.

  • eegstud24eegstud24 Williamsburg, VA
    edited May 2020

    Right. I know what an extension cable is. I cited Richard there more to illustrate my confusion about the primary purpose of using an extension cable. As for the batteries, I use the recommended one from OpenBCI: https://www.sparkfun.com/products/13851. I discarded the battery pack since I intended to use the rechargeable lithium ion batteries, but I could try my spare lithium ion battery?

  • wjcroftwjcroft Mount Shasta, CA

    It's possible your lithium single cell might be wonky. It only supplies 3.7V. The AA pack is 6V. The regulator on the mainboard is 3.3V. So margins are kind of close. A good fully charged lithium should be fine, but we're trying to rule out some issue with your power supply. The 6V AA pack would do that, with new alkaline cells.

  • eegstud24eegstud24 Williamsburg, VA

    I see, I'll try to get my hands on one from Amazon. Thank you two for your continued support. I'll check in once I get results.

  • wjcroftwjcroft Mount Shasta, CA

    You also said on an earlier post, "...I'll try the fixes Richard mentioned; I think my battery might in fact be getting really low. Thank you so much for the help you two...". Unless you have a more expensive charger, the charge level you are getting may also be suspect.

  • eegstud24eegstud24 Williamsburg, VA

    I'm having some trouble finding 6V AA battery packs with JST connectors. Any recommendations for purchasing one?

  • wjcroftwjcroft Mount Shasta, CA

    Ah, it looks like this is going to be harder than we thought since you no longer have the AA holder. I do see that Adafruit has both 4 cell AA holders and separately, JST-PH connectors attached to a short wire pair. But no holders like the one OpenBCI gave you. Similarly these are available on Amazon separately. If you had the JST-PH+wire-pair and a holder, you could splice them yourself.

    Here is another way you could eliminate "low battery" as a potential culprit. Do you have a DVM, digital voltmeter? Using the probes on that you can touch the + and minus (red and black) areas to measure the battery voltage, WHILE the Cyton is both powered on, streaming and dropping packets. Easy to access points are at the battery itself, under the transparent tape wrap, or behind the Cyton near the JST-PH connector where you can see the plus minus labels.

    With your battery fully charged and Cyton powered on / streaming to the laptop, you should see probably around 3.6 or 3.7 volts. On a freshly charged lithium, voltage can be as high as somewhat over 4 volts. All these values are fine as long as the voltage is over 3.3 volts, which where the voltage regulator on the Cyton is set.

    If your voltage looks ok during streaming, then it seems unlikely in my opinion, that the AA pack would give any better results. Cyton and Ultracortex is supposed to work with normal lithium cells.

    Also mention what charger you are using for the lithium cells, and how does it indicate full charge?

  • eegstud24eegstud24 Williamsburg, VA

    Okay great! I can get a DVM really quickly. I'll probably need it in the future anyway. I'm using the recommended one from OpenBCI: https://www.adafruit.com/product/1304. It indicates full charge with an LED, so I charged the battery for a couple of hours until I saw the green LED. I'll check back in with my voltmeter results in a few hours from now.

  • wjcroftwjcroft Mount Shasta, CA

    Remember, measure Cyton under load while streaming.

  • eegstud24eegstud24 Williamsburg, VA

    I measured the Cyton at the back where the + and - signs are. The multimeter returned about 4 V. One thing I do want to note is that packet drops are very rare when I'm not wearing the headset. As a result, I found it really hard to capture a moment where there was packet loss when I was using the multimeter with the headset off.

  • wjcroftwjcroft Mount Shasta, CA

    OK, just the fact that it was streaming at the time, means that the Cyton+RFduino was drawing the most power that it uses. So if the voltage was lower than required (3.3V), that would have shown up on your multimeter as you were streaming.

  • wjcroftwjcroft Mount Shasta, CA

    "One thing I do want to note is that packet drops are very rare when I'm not wearing the headset."

    Could this have some correlation with how close the Cyton mainboard is to the dongle? Did you ever clip the dongle to your shoulder or rear collar area (via extension cord and spring clip), while wearing the headset?

    Admittedly, your RFduino(s) sound like they have weak radios, which is the root of this.

  • eegstud24eegstud24 Williamsburg, VA

    I think you might be on to something. When I perch the dongle on my shoulder using the USB extension cable, all the packet drop problems go away. But the instant I put it back on my desk, they reappear again. I would think the RFduinos have better range than this?

Sign In or Register to comment.