Improving Signal Quality with Dry Electrodes
MrKlean518
Las Vegas
Hello, I wan to ask some questions (and possibly start a general discussion) about the best way to maximize the signal-to-noise ratio using a Cyton/Daisy with the dry electrodes. I have recently acquired a MindAffect decoder and when running their test script with the signal-quality test, I had a hard time every reaching an SNR below 12, when MA recommends getting a ration less than 5. So far, I have made sure to remove any wireless transmission devices around to prevent dropping samples. I have also replaced all of the electrodes with brand new dry-combs. I am also bald so getting adequate skin contact is not an issue. A few questions I have regarding the issue:
1. Whenever I put the headset on, the dry combs scrap against my head the whole way down. I am using the largest headset and the electrodes are barely screwed in yet it is still difficult to make adequate clearance. Is it possible that this is rapidly degrading the quality of the electrodes?
2. Do the ear-pieces have conductive coatings that can also degrade over time? I am wondering if I need to replace those.
3. Are there any better quality dry electrodes that have been used with the Ultracortex Mark IV?
While this is mostly examined while using the MindAffect decoder, I also notice a lot of noise when running SSVEP experiments independent from the MA system and so it is not just isolated to the decoder.
Comments
MrKlean, hi.
I have not seen the MindAffect instructions. Is there a link or pdf to read them?
SNR is an abbreviation for Signal to Noise Ratio. Generally higher means more signal, less noise. Perhaps MindAffect has some different ways to interpret this. See this post for typical impedances with the dry electrodes,
https://openbci.com/forum/index.php?p=/discussion/comment/13328/#Comment_13328
https://openbci.com/forum/index.php?p=/discussion/2517/expected-impedance-range-with-ultracortex-cyton-resolved
The 5mm dry combs wear out faster than the original 'spiky' combs. All the dry electrodes have a silver chloride coating which should appear dark grey. If that wears off there is a silver plate below that, and below that a plastic foundation. Silver chloride coating needs to be greyish, otherwise it has worn away.
Some noise effect may be due to mains (60 or 50 Hz) contamination. Which can be helped by positioning your setup away from wall warts, power supplies, extension cords, lighting fixtures, wall/floor conduits, etc.
Regards, William
SSVEP noise can also be due to jitter in your light source. This is especially the case with programmatically controlled monitor screen flashing. Several previous threads / posts addressing this.
I will try to run their program again and see what their metric was. It mentions the SNR but then it specifies what it actually uses to measure signal quality and what metric it displays and get back to you later today. I have a bit more time today in general, so I will try to do more to examine the different conditions of signal quality and write a more lengthy reply on it. I still have a few questions though.
1. How many uses should dry-combs get before degrading? I am trying to figure out if the tight fit of the electrodes which makes them rub against my head is making the coating degrade faster.
2. Do the ear pieces have coatings that can degrade as well?
3. Have any 3rd party dry-electrodes been tested with the Ultracortex M-IV?
I will also show the graphs from my SSVEP trials if I can find them so you can observe the type of noise I am looking at and see if it seems pretty standard all things considered.
https://www.fri-fl-shop.com/
Asked this previously: can you provide the instruction link for MindAffect. Do they state that dry electrodes work fine, or do they recommend paste or gel based? I thought I saw earlier pictures from them showing a plain looking 'headband' type arrangement. So using the entire Ultracortex headset may not be needed.
William
Please use the OpenBCI_GUI to obtain 'impedance' measures on each channel. I included on previous post, what range in ohms is acceptable.
The link for the Mind Affect wiki is here: https://github.com/mindaffect/General/wiki
It is originally configured for the Ganglion, but the wiki details how to configure the device for the Cyton with 8-channels and even says the normal configuration of electrodes for the Ultracortex can be used. In my e-mail correspondence with the people over at Mind Affect they have told me that you can rearrange the electrodes however you like to maximize signal quality, so I am working with a version I put together that localizes all 8 electrodes over the Parietal/Occipital lobes. In my correspondence, they also mentioned that they generally used gel-cup electrodes or water-based electrodes, but that the dry-electrodes CAN work, they just can be a little bit difficult to work with.
Will test impedance soon as well and share results.
Hey, nice find on the FRI gold combs, that is a new product.
https://www.fri-fl-shop.com/product/reusable-5mm-gold-spike-snap-electrode-tde-212b/
Unfortunately, the current combs and cones require a center hole for screw attachment. May be possible with a carefully aligned drill press to do a center hole. (Would not work with a hand drill.) Another consideration is what are these made of? What is the gold plated onto? Generally silver chloride electrodes are superior to all other types of electrode metals. Whatever gold plating exists on these, would be EXTREMELY thin, so would not guarantee longer lifetime. Dry contact gold electrodes would have to be tested to see how much more noise they produce than silver chloride.
As you mention, passive wet electrodes will always give better results than passive dry electrodes.
Impedance Update
Channels 1-8 have the following impedances; 700k, 400k, 2000k, 600k, 800-3000k (this one varies a bit), 400k, 30k, 450k. So other than electrode 3 (and sometimes 5) they seem to be in appropriate ranges. I will retest with the Mind Affect and see what happens there.
As mentioned, these posts discuss nominal ranges for the Ultracortex dry electrodes. Yours seem to be ok, with the possible exception of the 3000K, which is on the edge of being troublesome. Higher impedance electrodes are more subject to mains noise.
https://openbci.com/forum/index.php?p=/discussion/comment/13328/#Comment_13328
https://openbci.com/forum/index.php?p=/discussion/2517/expected-impedance-range-with-ultracortex-cyton-resolved
In addition to "Higher impedance electrodes are more subject to mains noise", we can also say that the microvolt signal level from passive dry electrodes will always be somewhat less than what is received from passive wet (gel or paste) electrodes. For some applications this may be important, which is why they say "...but that the dry-electrodes CAN work, they just can be a little bit difficult to work with..."
If you are getting significant mains noise (viewable on the FFT widget), then positioning your station away from all sources of EMF may improve that: power cords, wall warts, wall/floor conduits, extension cords, lighting fixtures, etc.
Okay so in regards to Mind Affect, they measure the Noise-to-Signal ratio, not the SNR. I have included a screenshot of their quality test so that you see the results I am getting. This screenshot actually happened to happen during arguably the best moment of the quality test. There were moments where the ratios went as high as 30-40. I am going to examine electrode 5 and see if there is an issue with that one.
Please try my suggestions regarding positioning away from mains noise. It would be instructive to see how they measure this. Can you ask them?
Also ask them if their 'NSR' ratio at all depends on room lighting, brightness of monitor screen, contrast, etc. With a cVEP setup, they are probably checking how strong a signal is received from the screen flashing. Turning down ambient room lighting, may make the contrast of the flashes, more apparent.
Hey WIlliam, sorry for the late reply, been busy with some other things so I haven't had the time to address these/test. I will likely get to that tomorrow morning.
For whatever reason I did not notice your response to my comment about the gold combs. I don't know the answer about their quality versus the silver chloride coated electrodes. However, the site does make the distinction that the silver chloride coated electrodes are disposable while they do not say the same about the gold combs. Additionally, they are considerably more expensive ($100 for 20) so I would hope that translates to longevity of use and/or signal quality. I will reach out to FRI and ask some questions and let you know what I hear. I know the mounting style is significantly different, but I was actually interested in designing a headset that is specifically contoured to my own head that I could inlay the snap-on cables and have the electrode position always fixed and easily replaceable (similar to the Neurosity notion design). Assuming I sort out the proper connectors for the wires to the board and work out the impedance it would probably work with some tweaking right?
Edit: Although, I suppose if I am designing it for myself and I am bald, it would be optimal to use all flat electrodes wouldnt it?
Jadin, hi.
re: 'disposable' vs. non-disposable electrodes. This is a complex subject area. Any type of electrode that has plating layers, will not last indefinitely. As the plating wears, layers underneath will be exposed. And these under layers are typically different metals that can form a battery 'galvanic' effect with the top layer metal.
Thus no gold electrodes are solid. They might be solid tin underneath, plated with gold. That is how most gold cups are made. Or silver underneath. Or even plastic base, first plated with tin or silver, then gold.
The only electrodes not subject to these galvanic effects are solid metal, same material throughout. So solid tin and solid silver are examples. These have the longest lifetime.
Most silver chloride type electrodes are labeled 'disposable'. Yet the FRI comb electrodes will last a long time when used dry with Ultracortex. If gel is used with them, the lifetime is shortened.
William