Cyton + Ultracortex RAILED Issue
severinchik
Kiyv
Hello, William @wjcroft and Richard @retiutut
Some time ago I received our Mark IV headset with 16 channels (Cyton & Daisy included), but I faced some problems using it.
I contacted to the OpenBCI support team, but they told that we need some outside opinion and advised to contact you regarding our issue.
After assembling Mark IV headset with Cyton & Daisy according to the instructions, I found out that amplitudes were higher that expected, also a lot of channels were RAILED when I 100% sure that they attached to the head.
OpenBCI guys decided that it was a hardware problem and provided me with the new Cyton. I tested it and faced some issue with signals behaviour - even when the helmet is placed on head all channels are RAILED.
May be you have some special experience with OpenBCI issues.
Here is the link to the folder with all my videos, they will help you with diagnosis: https://drive.google.com/drive/folders/1428a4icmSavNm3dxdYs-zxfPXovPUURM?usp=sharing
I'll be very grateful if you find time to check them. If you need some other videos for better understanding - just tell me.
Looking forward for your answer, Alexandra
Comments
You have come to the right place and tagged the right people. We will look into this for you! I might be busy this weekend, but letting you know I see this post, and I will investigate this soon.
@retiutut, thank you very much for such quick response
ok, ping me if you have some news
Alex, hi.
Please look over these threads,
https://openbci.com/forum/index.php?p=/discussion/1481/how-to-become-de-railed
https://www.google.com/search?as_q=railed&as_sitesearch=openbci.com
William
I took a quick look at the videos. Hmm, none of these seem to show the headset actually on your head. Without the scalp connection, of course everything looks railed. And just touching the pins is a very rough test, but no where near as valid as the actual test on your head. All EEG amplifiers go a bit crazy when pins are floating, because the amps are highly sensitive.
I'd also suggest you try a test with gold cups and paste if you have those.
How did the Cyton perform with the steps of the tutorial?
https://docs.openbci.com/docs/01GettingStarted/01-Boards/CytonGS#v-connect-yourself-to-openbci
See this thread for typical expected impedance values you can measure from the dry electrode headset. Somewhere in the range of say 1000K, 2000K, 3000K ohms. The lower the better, but even 2000K and 4000K can work, if your mains noise is not too severe.
https://openbci.com/forum/index.php?p=/discussion/2517/expected-impedance-range-with-ultracortex-cyton-resolved#latest
You turn on the impedance measurement per channel, by pressing the "ohm sign" button once, which is next to the channel number. Pressing the ohm again, turns off the impedance check.
Hello, guys
@wjcroft , I've checked the instructions you sent
I don't have the gold cups
I run some experiments and noticed that when Daisy connected to Cyton - all channels are railed, but when Daisy is disconnected the amplitude is huge. I added the video, maybe it'll give you some more information
https://drive.google.com/file/d/1LgoMrlUZahwTv1RFgE_tAjlGQILTDV93/view?usp=sharing
If not, I'll try fixing FTDI methods
Thank you
Alex, hi. It appears from this latest video, that you have some issues with your 50 or 60 Hz power line AC mains, causing interference. This is shown when your hands touch the table or laptop.
Do you have a different experience when the laptop is running off of battery, instead of plugged into the adapter? (Power adapter completely disconnected from laptop.) This could be a sign that there are some odd connections / ground loops happening with your adapter. Have you tried testing in other locations? There also may be power conduits running very close to your position. EMF electromagnetic fields, extend about 3 to 4 feet from conduits, extension cords, adapters, clocks, appliances, etc.
How tight are the electrodes on your head? You should be feeling the tips of the combs contacting your scalp.
Mentioning Richard @retiutut, if he might have further suggestions. If the battery test gives more promising results, and it looks like the noise is from the adapter, we can make further tips.
Regards, William
Did you see my earlier instructions about how to measure the impedance of your Cyton channels? That would be helpful. Also, try swapping the two ear clip connections. So that the clip used for Bias becomes the Reference, and vice versa. Since ALL channels are RAILING, this might indicate you have a bad ear clip reference impedance, which is common to all channels.
So no one in your lab or university or related fields, has some other types of electrodes you could test? Going through the steps of the Cyton tutorial measuring EEG, ECG, EMG, would provide helpful insights.
Lastly, another test you can do, if you have clip leads with spring jaws -- is to connect all three of these pins together: channel (you choose pin, pin closest to board), SRB2 (closest to board), Bias. With all these connected, the amplifier should measure close to zero microvolts, and very low impedance.
@severinchik Two things I recommend to debug this issue:
Try again using just Cyton, and share a screenshot of what the signals look like. Thank you for sharing the video with important details!
Hello, guys
@retiutut, @wjcroftx, I'm very grateful for your advice
I tried to connect both ear clips to the Cyton board, moved further from the laptop and amplitudes were near 30 uV:
https://drive.google.com/file/d/1M025D5PyWKB-8qe9cs-8xlgEyi7Kbrzc/view?usp=sharing
But when the Daisy is connected some channels are railed, even if they weren't railed while Daisy was disconnected:
I made several experiments, checked all connections and fitting electrodes to my scalp, but nothing helped
https://drive.google.com/file/d/1LizvSCmq1TR4JUQ1CCNRaiEOtIIreFws/view?usp=sharing
Please, check the videos attached
Do you have the 'Y-cable' connected between the SRB2's on both mainboard and Daisy? Y-cable then plugs into one of the ear clips for your combined reference.
WITHOUT the Daisy, please check your impedances as I asked twice previously. Did you also switch the two ear clips?
The important detail here is that the earclips are attached to the boards differently when using Cyton or Cyton+Daisy.
Also, it appears you have a wire connected to the AGND pin? You should only be connecting to SRB2 and Bias, never connect to AGND.
Hello, guys
@wjcroft, @retiutut
On this video I mistakenly connected the second part of the y-cable to the AGND
https://drive.google.com/file/d/1M025D5PyWKB-8qe9cs-8xlgEyi7Kbrzc/view?usp=sharing
But with the correct configuration I just got all channels railed
So I cannot measure the impedances.
https://drive.google.com/file/d/1MHCUsIJ_7oc-jiGnAuSS-_6FZ1eUZVCp/view?usp=sharing
The same situation this time I had with Cyton + Daisy - all railed channels
So, I don't understand why behaviour of boards is completely different from one experiment to another
Here is my wires configuration for Cyton + Daisy:
https://drive.google.com/file/d/1MLh8GB_yJ8siVwYDm7V7pmpgq9RlLYyi/view?usp=sharing
Let's forget about the Daisy, and just stick to the mainboard with 8 channels, until we figure this out the RAILED issue.
I do see from the 2nd link you gave, that your 8 channel connection to the mainboard alone, looks ok. The previous 8 channel video you posted (the first link) was getting better (not railing) signals with the 8 channels. I don't understand what is different now you are RAILING with 8 channels?
Have you tried swapping the two ear clips, in case one is damaged? If the reference ear clip is not a good connection (damaged wire, connector, ear clip, silver chloride coating, etc.) -- then all channels will be terrible. This is why I suggest swapping with the Bias/Ground ear clip to see if the clip/wire/connector is an issue.
If your technical crew has a DVM (digital volt meter), they can do a resistance check of each ear clip, measured from the connector to the clip. Should be very close to zero ohms.
I also mentioned another test you could do with one or more channels, by connecting the channel(s), SRB2, and Bias together using clip leads of some sort. This should show close to zero microvolts on your time series. If that succeeds, it proves that the mainboard is fine, which I believe you will verify. This issue is then something going wrong with the ear clips.
William
Hello, guys
I run experiments with Cyton alone and this time channels weren't railed.
I also tried to switch earclips, so here is initial config performance: https://drive.google.com/file/d/1NoWyUE8mqAwLSxwJgbHSiAaoxmG8KVYM/view?usp=sharing
and with switched earclips: https://drive.google.com/file/d/1jxH7gTHmVUD5uWGA_doZShq1P0n-3zyt/view?usp=sharing
Seems like earclips are OK
Here is recording of experiment with initial configuration of earclips (I moved ~3 meters from all active electric devices):
https://drive.google.com/file/d/14FagR4vRXFnr3XpL7E7rq0exde1apNMx/view?usp=sharing
P.S. I don't have a technical team or any additional stuff - only items included in Mark IV kit
Thank you, Alexandra
Can you just summarize so I don't have to watch the videos? Are you saying that Cyton mainboard alone works fine?
First 2 are pictures, and the last one is ~one minute screenrecording
Now the board works, but amplitudes are high even when I'm far from the electric devices
Did you check your impedances?
Impendances were in range ~100-600 kOhm
Also open up the FFT wider frequency range to see if you can discern other peaks.
Ignore my previous comment about the 32 Hz peak, that is just the injection used to measure impedance. Still, the 1st image shows a huge amount of wideband noise. Adjust the FFT for wider frequency coverage, to see if there is something missing.
The photo showing the 'switched' ear clips, has more noise (on channels 5-8) than the 'non-switched' setup.
Your recording on the 3rd link is helpful for checking impedances, but cannot be used to differentiate between multiple locations in the rooms (as having less noise). This is because the 31, 32 Hz injection for impedance measurement, swamps out all the other signals. Impedance check must be turned off for checking actual noise levels.
Please adjust the range of the FFT so we can see the source of the high frequency noise. I assume you have your notch filter set correctly? (To your AC mains frequency). This noise sure looks like it could be mains related.
Hi @wjcroft!
I adjusted the range of the FFT as you asked and turned off impedances measurement
Here is the result (~50 seconds recording)
I started the data stream, then moved further from the laptop and then came closer to stop recording - you'll see it by the amplitudes changing
When I'm standing far from the electric field the amplitudes seems to be in the normal range, am I right?
Thank you)
I'm sorry this is taking so long to debug. It's unclear to me what your video shows. I did ask previously if you could expand the frequency range of the FFT, so that it includes higher frequencies. Up to and beyond the mains frequency. Which is likely 50 Hz there? It's 60 Hz here in the US. I also asked about your notch filter, it is possibly set wrong. The notch filter must be turned on and set at the mains AC frequency to block out more of that frequency. The mains EMF electromagnetic field is so strong in any building, that the digital signal processing needs to 'notch' it out, otherwise it overwhelms the normal EEG.
It looks to me from all your videos that you have some serious mains interference going on. The EMF from extensions cords, power supplies, wall / floor / ceiling conduits, fluorescent lighting, etc. -- extends out several feet in all directions from the source.
https://docs.openbci.com/docs/10Troubleshooting/minimizingNoise#get-rid-of-ac-noise
I have made an issue on our Documentation repository to clearly label pictures of the Cyton board to show how Earclips should be connected to the board. This was hard for me to figure out when I got started, and this needs to be super obvious for new users.
https://github.com/OpenBCI/Documentation/issues/19