5 Hz Noise/Artifact in all 16 channels [resolved]
Hi
I am using the Cyton 16 channel (daisy chained) board and the electrode cap for my experiments. I am just getting started and trying to make sure I am getting what I am supposed to get. I have tried both ways: with and without the gel. In all trials I am getting a peak at around 5 Hz for almost all the channels in the FFT plot. Is this common/correct? In most EEG waveform FFT plots I have seen on this forum, there is no such peak. So I was wondering what is causing this? How should I go about solving the issue. I am attaching a snapshot of the fft plot here: https://drive.google.com/file/d/1HMzs8LUZT1IgBFOXz5PK1G6iUC_xceQZ/view?usp=sharing
Also note that I have tried the setup on three different subjects (including me) and in all cases there was this peak. Another thing I would like to add is I have done the open-close eye alpha wave test and I could notice visible increment in the alpha wave power when eyes were closed.
Comments
Max, hi.
Gel is necessary with elastic electrode caps. Did you check your impedances, by clicking the ohm symbol on each channel? (Ω)
Where is your reference, is it connected to SRB2? If you expand the FFT scale to view the mains frequency (50 or 60 Hz), do you see mains noise? Mains noise can couple into the EEG, if your chair / table is near power supplies, conduits, lamps, extension cords, power strips, etc. Try moving to another location and see if it changes.
One of channels in grey, has much more of this noise than the others. Why is that? Impedance of that channel vs the others?
Regards, William
Hi William
Thanks a lot for your prompt response!!
This is today’s result from my old seating position: https://drive.google.com/file/d/13Z47HUkAxeAvUlVydx_a3ZBTcOaX5M6W/view?usp=sharing
But then I moved to a different room with lesser power chords and then I got this:
https://drive.google.com/file/d/1sTL3X1ixKasuSlySC4IaNGtvjId8_JQ8/view?usp=sharing
Does this seem right now?
Thanks
Tinish
Yes, much better. As you can see from your first photo, BOTH 60 Hz mains and 5-6 Hz cross modulation are present. But it is the mains that is the root cause.
Still recommend checking your impedances, to see if they all look reasonable.
Hi William
Thanks!!
I checked the impedances of all the channels. They were mostly between 11k to 25 k ohms. There were a few channels which were > 40k ohms, but I was able to bring them down to below 25 k ohms by injecting more gel in their slots. Is this range good enough?
Thanks
Tinish
Yes, that's fine. The impedance check is mainly to discover 'outliers' in terms of being substantially above the rest of the electrodes. That is then a sign they need some attention.
Great!! Thanks a lot for your help in solving this issue.
For more background on skin impedance vs. amplifier INPUT impedance (1 gigaohm in the case of Cyton), see this post,
https://openbci.com/forum/index.php?p=/discussion/comment/13131/#Comment_13131
Hi William!
I have also found such cross modulation in my recordings, but while simultaneously recording from another amplifier, I did not find such cross modulation even with the presence of mains noise. I do not have much electrical knowledge, so can you please explain why such cross modulation occurs? I have attached a paper here that seems to explain this, but I am unsure if it is relevant.
https://www.kaajakari.net/~ville/research/publications/kaajakari05uffc.pdf
Hi Flying,
How does your FFT and time series look inside the GUI? Do you have a notch filter at your mains frequency (which appears to be 50 Hz)? The other 'harmonics' mirroring the mains will likely go down once you employ the notch. Your other recording system may well automatically employ such a notch at mains.
William
Hi William,
I do not want to use an online notch filter. I had switched off the notch filter in the other system as well. I get these same cross modulation bands when I use OpenBCI GUI. I just wanted to understand why such cross modulation occurs.
A notch filter is pretty much standard in most EEG applications. The 50 Hz mains frequency is SO HUGE compared to the microvolt level EEG, that it causes cross-modulation, intermodulation distortions, and those are the other peaks you are seeing.
https://www.google.com/search?q=cross+modulation+vs+intermodulation
https://www.rfwireless-world.com/Terminology/cross-modulation-vs-intermodulation.html
https://en.wikipedia.org/wiki/Intermodulation
William
Okay, but aren't the online filters just digital filters done at the last step? So would an offline notch filter also reduce these cross modulation bands? I was under the assumption that this amplitude modulation occurs at the hardware level.
A proper notch filter at the mains frequency will NOT distort your EEG signal. Almost all the EEG activity is below 50 Hz. There is EEG near / above 50 Hz, and near 40 Hz (typically referred to as Gamma), but the skin and scalp and EMG activity of scalp muscles, greatly diminishes any ability to reliably utilize it.
https://www.google.com/search?q=gamma+eeg+scalp+contamination
If the extra bands disappear when you have a notch at 50 Hz, then they are artifacts of the mains. I'm not sure how exactly the harmonics / subharmonics are spawned by the 50 Hz, it's a complex phenomena related to modulation products and non-linearity.
William
Okay, thank you!