Electrical Grounding problem
April 22, 2011 8:58 AM Subscribe
How to ground (shield?) a laptop to get a good EEG signal?
I am learning to do neurofeedback, and am having trouble getting a good EEG (electroencephalographic) signal on two of my three laptops. One works fine, but it's wonky and may not be long for this world, so I must get at least one of the others to work.
The problem is this: there seems to be some sort of electrical interference on both computers that makes it impossible to get a clean EEG signal. I have a spectrograph on my software (the software is called BioExplorer) which lets me see what waveforms are prominent, and what I see is, surprisingly, there is little 60-cycle, which would be the most obvious culprit. Instead, there is a spike at the very low end of the spectrum, just greater than 0 Hz.
Here is a screenshot:
You can see the spectrograph all colorful on the bottom, and the raw EEG signal is on top. It should look like nice waves and not be all spiky like that.
I am using the laptops on battery power, because I've been advised that that's the most electrically clean way of doing it, but now somebody else on a neurofeedback list I'm on is telling me to plug it in to a three-prong adapter and into the wall instead. He is also telling me to remove a screw from an old laptop (the type of screw that's used to screw a serial port thing into the serial port, that we don't see so much anymore because we use USB instead) and screw that screw into -- something --- I can't figure out what ("any port that has a hole for a screw" - what??) --- the point is, he's trying to tell me how to make some sort of antenna or ground or something.
If you know anything about what I'm talking about here: why I can't get a clean signal from relatively new laptops, why they might not be "shielded" properly, and what I can do about it, I would be very grateful!
This person (a neurofeedback veteran) has also directed me to this site where they sell all sorts of grounding stuff, but I have no idea where to start looking for something that would help.
I am learning to do neurofeedback, and am having trouble getting a good EEG (electroencephalographic) signal on two of my three laptops. One works fine, but it's wonky and may not be long for this world, so I must get at least one of the others to work.
The problem is this: there seems to be some sort of electrical interference on both computers that makes it impossible to get a clean EEG signal. I have a spectrograph on my software (the software is called BioExplorer) which lets me see what waveforms are prominent, and what I see is, surprisingly, there is little 60-cycle, which would be the most obvious culprit. Instead, there is a spike at the very low end of the spectrum, just greater than 0 Hz.
Here is a screenshot:
You can see the spectrograph all colorful on the bottom, and the raw EEG signal is on top. It should look like nice waves and not be all spiky like that.
I am using the laptops on battery power, because I've been advised that that's the most electrically clean way of doing it, but now somebody else on a neurofeedback list I'm on is telling me to plug it in to a three-prong adapter and into the wall instead. He is also telling me to remove a screw from an old laptop (the type of screw that's used to screw a serial port thing into the serial port, that we don't see so much anymore because we use USB instead) and screw that screw into -- something --- I can't figure out what ("any port that has a hole for a screw" - what??) --- the point is, he's trying to tell me how to make some sort of antenna or ground or something.
If you know anything about what I'm talking about here: why I can't get a clean signal from relatively new laptops, why they might not be "shielded" properly, and what I can do about it, I would be very grateful!
This person (a neurofeedback veteran) has also directed me to this site where they sell all sorts of grounding stuff, but I have no idea where to start looking for something that would help.
I work with human scalp EEG in a laboratory setting. Forgive me if I'm telling you anything you already know.
Low-frequency (less than 1 Hz) artifacts = grounding problem. If you plug it into the wall, you'll be grounded, but you'll also probably introduce 60 Hz from your AC/DC adapter (aka power brick). You could have the best of both worlds by disassembling a 3-prong plug and running a wire from the grounding pin to somewhere on your laptop chassis without connecting the actual "power" parts of the plug. Do at your own risk-- AC electricity is dangerous.
HOWEVER. The "spikiness" is not caused by low frequency components. Sharp changes in voltage are necessarily high frequency. The good news is that the most common high-frequency artifact is 60 Hz power-line noise, and that's not what I'm seeing. The second most common is muscle artifact, but that would overwhelm any neural signal.
Honestly, your EEG looks pretty good to me, for raw, unfiltered EEG. (It is unfiltered, right?) You're getting good enough signal to sort of see alpha waves. If you want it to be smooth waves, you need to filter it. I know very little about neurofeedback; what frequency ranges are you interested in? Theta, beta, alpha, gamma? If you don't care about gamma, and you want to get rid of the low frequency components, just bandpass your signal from 1-20 Hz.
If you do care about gamma, well... good luck. High-frequency components of EEG are necessarily lower in amplitude than low-frequncy components. It takes sensitive electrodes and amps to pick that stuff up in a laboratory setting, where we're specifically interested in gamma. (Our EEG rig, is a $100k piece of equipment, though granted, it's 128-channel.) That's why clinical and less-intense cognitive EEG applications tend to be focused on lower frequencies, usually with an eye towards ERP components, rather than total spectral decomposition.
Just out of curiosity, what is your sampling rate?
posted by supercres at 9:20 AM on April 22, 2011 [1 favorite]
Low-frequency (less than 1 Hz) artifacts = grounding problem. If you plug it into the wall, you'll be grounded, but you'll also probably introduce 60 Hz from your AC/DC adapter (aka power brick). You could have the best of both worlds by disassembling a 3-prong plug and running a wire from the grounding pin to somewhere on your laptop chassis without connecting the actual "power" parts of the plug. Do at your own risk-- AC electricity is dangerous.
HOWEVER. The "spikiness" is not caused by low frequency components. Sharp changes in voltage are necessarily high frequency. The good news is that the most common high-frequency artifact is 60 Hz power-line noise, and that's not what I'm seeing. The second most common is muscle artifact, but that would overwhelm any neural signal.
Honestly, your EEG looks pretty good to me, for raw, unfiltered EEG. (It is unfiltered, right?) You're getting good enough signal to sort of see alpha waves. If you want it to be smooth waves, you need to filter it. I know very little about neurofeedback; what frequency ranges are you interested in? Theta, beta, alpha, gamma? If you don't care about gamma, and you want to get rid of the low frequency components, just bandpass your signal from 1-20 Hz.
If you do care about gamma, well... good luck. High-frequency components of EEG are necessarily lower in amplitude than low-frequncy components. It takes sensitive electrodes and amps to pick that stuff up in a laboratory setting, where we're specifically interested in gamma. (Our EEG rig, is a $100k piece of equipment, though granted, it's 128-channel.) That's why clinical and less-intense cognitive EEG applications tend to be focused on lower frequencies, usually with an eye towards ERP components, rather than total spectral decomposition.
Just out of curiosity, what is your sampling rate?
posted by supercres at 9:20 AM on April 22, 2011 [1 favorite]
Can you post a screen shot of the EEG from the computer that does work well? I'd like to see what you're hoping to achieve. Also please check any differences in settings, especially sampling rate and filtering.
posted by supercres at 9:23 AM on April 22, 2011
posted by supercres at 9:23 AM on April 22, 2011
Every laptop connector I've bothered to dig into has had only plus and minus coming out of the power supply, even when the power supply has three prongs. Internally, the laptop uses that minus as its ground. So plugging it in (especially with a modern switching power supply which doesn't have an isolating transformer) is more likely to cause noise than running on batteries.
Usually with EEG stuff you have ear clips that are used as the body's reference ground. However, as adipocere points out, that whole subsystem should run off of batteries and be optically isolated from the portions that talk to the computer.
And everything that supercres said: That looks like what I've seen out of various different bits of hardware, EEGs are noisy, and the spiky bits could be muscle movement, bad contact from the electrodes, all sorts of stuff.
Can you tell us a little more about your hardware setup, sampling rates, etc? Are you seeing this noise when you touch the laptop with the probes on? Since the sampling device should be isolated, is there metal on the sampling device you can touch? Does that change things?
posted by straw at 9:24 AM on April 22, 2011
Usually with EEG stuff you have ear clips that are used as the body's reference ground. However, as adipocere points out, that whole subsystem should run off of batteries and be optically isolated from the portions that talk to the computer.
And everything that supercres said: That looks like what I've seen out of various different bits of hardware, EEGs are noisy, and the spiky bits could be muscle movement, bad contact from the electrodes, all sorts of stuff.
Can you tell us a little more about your hardware setup, sampling rates, etc? Are you seeing this noise when you touch the laptop with the probes on? Since the sampling device should be isolated, is there metal on the sampling device you can touch? Does that change things?
posted by straw at 9:24 AM on April 22, 2011
(Whoops: Additional Q: Are your ear clips properly seated and gel'd?)
posted by straw at 9:24 AM on April 22, 2011
posted by straw at 9:24 AM on April 22, 2011
That's a DC offset. You can see it in your waveform, which is not centered at zero. The first bucket of the FFT is the 0Hz (DC) component.
The guy isn't telling you to actually power the laptop from the wall, rather to use the ground pin of the outlet to ground the laptop, otherwise it's at a floating potential. One way of doing that is to take a standard 3 pin power cord and connect the ground pin to any metallic part of the laptop, leaving the actual hot/neutral pins unconnected. You could also just do the same with a regular piece of wire if you have metal pipes that go underground.
I'm not sure grounding is really the issue here though. On the laptop that works properly, are the vertical divisions about the same (i.e. 10 uV/div)? By that I mean is the output in the same general range? If you run an amplifier with very low inputs then the natural DC offset of the instrumentation op-amp will contribute. Is this thing powered entirely by a COM port, or something else? If so, then check the rails of the COM port with a DMM to see if they are stable and correct. Often modern computers skimp on proper COM/LPT voltages because for signaling it doesn't really matter if you don't have the full 12V, but if this thing expects to draw power from a COM port then it will matter. That is one area where older hardware generally is better.
posted by Rhomboid at 9:24 AM on April 22, 2011
The guy isn't telling you to actually power the laptop from the wall, rather to use the ground pin of the outlet to ground the laptop, otherwise it's at a floating potential. One way of doing that is to take a standard 3 pin power cord and connect the ground pin to any metallic part of the laptop, leaving the actual hot/neutral pins unconnected. You could also just do the same with a regular piece of wire if you have metal pipes that go underground.
I'm not sure grounding is really the issue here though. On the laptop that works properly, are the vertical divisions about the same (i.e. 10 uV/div)? By that I mean is the output in the same general range? If you run an amplifier with very low inputs then the natural DC offset of the instrumentation op-amp will contribute. Is this thing powered entirely by a COM port, or something else? If so, then check the rails of the COM port with a DMM to see if they are stable and correct. Often modern computers skimp on proper COM/LPT voltages because for signaling it doesn't really matter if you don't have the full 12V, but if this thing expects to draw power from a COM port then it will matter. That is one area where older hardware generally is better.
posted by Rhomboid at 9:24 AM on April 22, 2011
Agreeing with supercres that your waveform looks OK: the amplitude is in the generally expected range, no obvious line noise, no movement artifact (at least on this particular stretch of the recording). I'm not sure the DC offset is really going to change the overall picture in the frequency domain, at least not in the bands you're interested in for neurofeedback purposes (alpha? theta?).
posted by Nomyte at 9:59 AM on April 22, 2011
posted by Nomyte at 9:59 AM on April 22, 2011
Response by poster: Thank you for these great responses. Now I just have to take a course in Electricalese as a Second Language.
I'm going to try to find out the sampling rate and what optical isolation means.
Here is a screenshot of a really bad signal that's also typical of what I'm getting. I didn't show it before because it doesn't show 60-cycle -- but there wasn't much.
and
Here is a screenshot of a much better signal, obtained from my "good" computer.
posted by DMelanogaster at 10:39 AM on April 22, 2011
I'm going to try to find out the sampling rate and what optical isolation means.
Here is a screenshot of a really bad signal that's also typical of what I'm getting. I didn't show it before because it doesn't show 60-cycle -- but there wasn't much.
and
Here is a screenshot of a much better signal, obtained from my "good" computer.
posted by DMelanogaster at 10:39 AM on April 22, 2011
So, two things:
1. What's your sampling hardware? Are you using the microphone port?
2. You did notice that on your laptop scale each tick is 6.0, and on your '"good" computer' each tick is 30.0, right? Is the difference simply that you're viewing it on scales that differ by 5x?
posted by straw at 10:47 AM on April 22, 2011 [1 favorite]
1. What's your sampling hardware? Are you using the microphone port?
2. You did notice that on your laptop scale each tick is 6.0, and on your '"good" computer' each tick is 30.0, right? Is the difference simply that you're viewing it on scales that differ by 5x?
posted by straw at 10:47 AM on April 22, 2011 [1 favorite]
Response by poster: Oh, and also, here's the cute little EEG amp I'm using:
http://www.brain-trainer.com/cgi-bin/shop.pl?shop=get_item&item_id=11
Before you laugh, thousands of people (well, hundreds? no, probably a couple of thousand) are using these to do neurofeedback training with no problems (even *I* don't have a problem with my one moribund laptop).
The thing has a "dongle" that plugs via USB into the laptop, and the pendant and the dongle communicate wirelessly.
aha: "•Sampling rate: 122/128/256/512 sps"
and I am using earclips (1 ear for ground, 1 for reference) which, again, produce a clean signal on one laptop.
I have also prepped the ears and head with a prep paste and alcohol and used EEG paste on earclips as well as CZ. The subject is my husband and, perhaps tragically, hair is not a problem in this case.
posted by DMelanogaster at 10:53 AM on April 22, 2011 [1 favorite]
http://www.brain-trainer.com/cgi-bin/shop.pl?shop=get_item&item_id=11
Before you laugh, thousands of people (well, hundreds? no, probably a couple of thousand) are using these to do neurofeedback training with no problems (even *I* don't have a problem with my one moribund laptop).
The thing has a "dongle" that plugs via USB into the laptop, and the pendant and the dongle communicate wirelessly.
aha: "•Sampling rate: 122/128/256/512 sps"
and I am using earclips (1 ear for ground, 1 for reference) which, again, produce a clean signal on one laptop.
I have also prepped the ears and head with a prep paste and alcohol and used EEG paste on earclips as well as CZ. The subject is my husband and, perhaps tragically, hair is not a problem in this case.
posted by DMelanogaster at 10:53 AM on April 22, 2011 [1 favorite]
Response by poster: Sorry I forgot to make the url a link. It's here.
posted by DMelanogaster at 10:54 AM on April 22, 2011
posted by DMelanogaster at 10:54 AM on April 22, 2011
The device is wireless. This strongly suggests that all the sampling is being done in the pendant and is transmitting sampled data digitally to your computer. Which means that any differences between the two computers aren't due to grounding.
I strongly suspect that the problem is that the scales are set so that you're seeing 5x the scale on the laptop as the desktop.
On the hair issue: In response to growing baldness, I shaved my head. My next session, that little stubble rather than having the hair bend over caused me all sorts of contact issues. But, clearly, this isn't you're problem as you're just switching the receiver from one computer to another.
posted by straw at 11:02 AM on April 22, 2011
I strongly suspect that the problem is that the scales are set so that you're seeing 5x the scale on the laptop as the desktop.
On the hair issue: In response to growing baldness, I shaved my head. My next session, that little stubble rather than having the hair bend over caused me all sorts of contact issues. But, clearly, this isn't you're problem as you're just switching the receiver from one computer to another.
posted by straw at 11:02 AM on April 22, 2011
All laptops use DC-to-DC converters, to change the single voltage delivered by their batteries, into different voltages required by their logic circuits and internal devices (typically +12VDC, +5VDC and -12VDC). Depending on their design and quality, these devices can generate considerable electrical noise artifacts themselves, usually at multiples of 60Hz, where their internal "chopper" circuits operate (120hz, 240Hz, 480Hz, etc.). Modern motherboards have dozens of small decoupling capacitors all around the board, near each logic chip, to soak up this noise locally, but it can still exist and be promulgated to external devices on the USB chain, and by short range direct EMI, which normally also contain decoupling capacitors to filter it. So, it's not generally a problem in normal uses for laptops, and some laptops, by design, will produce more of it than do others.
You might try adding some clip-on style ferrite beads on your leads, very near the laptop, and on USB cables, to try to suppress this.
"The device is wireless. ..."
posted by straw at 2:02 PM
Generally, I agree with straw on this, but the device description also says "•Retractable USB extension cord, to ensure line of sight between pendant and receiver ", so if there is any EMI on the USB line, could that be a good antenna for noise? Might be worth a clip on ferrite bead, very near the USB port of the laptop...
posted by paulsc at 11:10 AM on April 22, 2011
You might try adding some clip-on style ferrite beads on your leads, very near the laptop, and on USB cables, to try to suppress this.
"The device is wireless. ..."
posted by straw at 2:02 PM
Generally, I agree with straw on this, but the device description also says "•Retractable USB extension cord, to ensure line of sight between pendant and receiver ", so if there is any EMI on the USB line, could that be a good antenna for noise? Might be worth a clip on ferrite bead, very near the USB port of the laptop...
posted by paulsc at 11:10 AM on April 22, 2011
If there's noise on the USB cable between the receiver and the computer, that's going to cause USB protocol errors. Those will most likely show up as much bigger glitches (ie: half-second dropouts, that sort of thing), and would also likely be found in glitching of other USB devices. The "line of sight" thing suggests that it's using IR (infra-red, light just below the visible spectrum, like your TV remote) for transmission (which, given the earlier comments about optical decoupling makes a lot of sense), further reducing the likelyhood that this is due to power supply noise or grounding issues.
My bet's still on misinterpreting the differing scales.
posted by straw at 11:28 AM on April 22, 2011
My bet's still on misinterpreting the differing scales.
posted by straw at 11:28 AM on April 22, 2011
If the wireless pendant is being used, that pendant has already sampled and digitized the analog signal. The data in digital form is then transmitted over to the USB dongle. So, there is no way electrical interference on the laptop or dongle could be causing the noise. The packets transmitted from the pendant to the dongle could suffer wireless interference, but any half-decent wireless protocol would have some error detection built-in, so that is less likely the issue as well.
With the exact same pendant, electrodes and USB dongle, are you seeing the problem with newer laptop, but not the older one?
posted by thewildgreen at 12:20 PM on April 22, 2011
With the exact same pendant, electrodes and USB dongle, are you seeing the problem with newer laptop, but not the older one?
posted by thewildgreen at 12:20 PM on April 22, 2011
Yeah, it's a voltage-scale issue. Your EEG is the same on both computers. Not great, but probably okay. If you want it to look pretty, bandpass- or lowpass-filter it before display.
USB is a digital signal. Once it gets there, EM interference isn't a problem anymore. You only need to be concerned about the signal while it's still analog, which is entirely in your EEG amp from the looks of it.
The reason you're not seeing 60 Hz noise is that your EEG amp simply doesn't process signals that fast-- it says it only goes up to 56 Hz. So you don't need to worry about it. You need a lowpass filter around 50 Hz (40 to be conservative) so you're not just seeing noise-- anything faster than that is not coming from your subject; it's artifact from your recording setup. At this point, you should ignore anything above 56 Hz, and if you filter it, you should ignore anything above your cutoff frequency. Your sampling frequency doesn't need to be much more than 100 if you're getting reliable signal above 50 Hz.
posted by supercres at 12:22 PM on April 22, 2011 [1 favorite]
USB is a digital signal. Once it gets there, EM interference isn't a problem anymore. You only need to be concerned about the signal while it's still analog, which is entirely in your EEG amp from the looks of it.
The reason you're not seeing 60 Hz noise is that your EEG amp simply doesn't process signals that fast-- it says it only goes up to 56 Hz. So you don't need to worry about it. You need a lowpass filter around 50 Hz (40 to be conservative) so you're not just seeing noise-- anything faster than that is not coming from your subject; it's artifact from your recording setup. At this point, you should ignore anything above 56 Hz, and if you filter it, you should ignore anything above your cutoff frequency. Your sampling frequency doesn't need to be much more than 100 if you're getting reliable signal above 50 Hz.
posted by supercres at 12:22 PM on April 22, 2011 [1 favorite]
And re: baldness-- bald subjects often have terrible impedance (which is what you're trying to minimize; low impedance means good scalp contact). The reason for that is that the skin is constantly exposed, and often exposed directly to the sun, leading to a layer of dead skin, further insulating the electrode from the brain.
If you want to get around that, unfortunately you have to abrade the skin with wipes like this, or by using an abrasive gel like this. I suspect that your EEG system is not designed to work at high impedance, so the cleaner the contact, the cleaner the EEG.
That said, if you don't care about anything above 20 Hz, you really don't have to worry about it. Filter everything above that out and carry on in the slower bands.
posted by supercres at 12:31 PM on April 22, 2011
If you want to get around that, unfortunately you have to abrade the skin with wipes like this, or by using an abrasive gel like this. I suspect that your EEG system is not designed to work at high impedance, so the cleaner the contact, the cleaner the EEG.
That said, if you don't care about anything above 20 Hz, you really don't have to worry about it. Filter everything above that out and carry on in the slower bands.
posted by supercres at 12:31 PM on April 22, 2011
Gah, should read:
Your sampling frequency doesn't need to be much more than 100 if you're not getting reliable signal above 50 Hz.
Which you're not. I don't know why they would include a digitizer that goes up to 512 s/sec when they admit that the system tops out at 56 Hz.
Feel free to memail if you like.
posted by supercres at 12:40 PM on April 22, 2011
Your sampling frequency doesn't need to be much more than 100 if you're not getting reliable signal above 50 Hz.
Which you're not. I don't know why they would include a digitizer that goes up to 512 s/sec when they admit that the system tops out at 56 Hz.
Feel free to memail if you like.
posted by supercres at 12:40 PM on April 22, 2011
Yep, I agree that the difference is just a scaling problem. Those two screenshots you show above are just about the same quality. You just have the scales set differently. Look at the vertical scale on the top left of each screenshot where it shows the raw waveform. One goes from 0 to 120u and the other goes from 0 to 24u. Effectively you have just zoomed out so that the noise looks smaller. If you adjust the scale for both to be the same, you should see similar results.
posted by JackFlash at 1:30 PM on April 22, 2011
posted by JackFlash at 1:30 PM on April 22, 2011
Response by poster: Very interesting, all of this!!
to answer the question:
"With the exact same pendant, electrodes and USB dongle, are you seeing the problem with newer laptop, but not the older one?"
Good signal on Toshiba R100 which is about 5 years old.
"bad" (although now we're questioning this) on IBM T42, several years ago, bought reconditioned so not sure of age, and on Toshiba I bought six months ago, pretty new.
So the oldest laptop has the best signal. They're all laptops, no desktop.
I am using NuPrep on scalp and ears.
I'm on a list of neurofeedbackers and they're also now telling me they don't necessarily think my signal is bad. Funny they're not talking about filters.
The biggest thing I see on the spectrograph is spiking between 0 and 2, which I'm told is "artifact" (muscle?), NOT 60-cycle.
I am only interested in frequencies between around 3 I guess and 20-something, that is, between theta and high beta.
The scaling issue is fascinating, thank you!
The consensus on the neurofeedback listserv is that the signal isn't bad, but they're asking me what my impedances are, which I did not check (because I don't have an impedance meter. I used to work with another machine which has one built in, but then I switched to this pendant. I can plug my electrodes into the old EEG machine (ProComp, by Thought Technology to check impedancs when I hook up again)).
Another person said that I might use the tops of the ears rather than the earlobes for my ground and reference (the clips) because they don't reflect the pulse as much (not that we know that the pulse is a problem).
Thank you for all of this feedback.
posted by DMelanogaster at 6:15 PM on April 22, 2011
to answer the question:
"With the exact same pendant, electrodes and USB dongle, are you seeing the problem with newer laptop, but not the older one?"
Good signal on Toshiba R100 which is about 5 years old.
"bad" (although now we're questioning this) on IBM T42, several years ago, bought reconditioned so not sure of age, and on Toshiba I bought six months ago, pretty new.
So the oldest laptop has the best signal. They're all laptops, no desktop.
I am using NuPrep on scalp and ears.
I'm on a list of neurofeedbackers and they're also now telling me they don't necessarily think my signal is bad. Funny they're not talking about filters.
The biggest thing I see on the spectrograph is spiking between 0 and 2, which I'm told is "artifact" (muscle?), NOT 60-cycle.
I am only interested in frequencies between around 3 I guess and 20-something, that is, between theta and high beta.
The scaling issue is fascinating, thank you!
The consensus on the neurofeedback listserv is that the signal isn't bad, but they're asking me what my impedances are, which I did not check (because I don't have an impedance meter. I used to work with another machine which has one built in, but then I switched to this pendant. I can plug my electrodes into the old EEG machine (ProComp, by Thought Technology to check impedancs when I hook up again)).
Another person said that I might use the tops of the ears rather than the earlobes for my ground and reference (the clips) because they don't reflect the pulse as much (not that we know that the pulse is a problem).
Thank you for all of this feedback.
posted by DMelanogaster at 6:15 PM on April 22, 2011
So if you've convinced yourself that there's more than the scaling issue, and you think the artifacts you want to eliminate are muscle twitch related, then I'd start to actually look at the physical environment for differences. Are the two laptops at the same height? Are the chairs identical? Lighting (especially glare off of one screen vs the other)?
This sort of debugging is a pain in the tuchus, but when you get out there on the hairy edge of technology it's what happens. Welcome to the frontier!
posted by straw at 9:09 PM on April 22, 2011
This sort of debugging is a pain in the tuchus, but when you get out there on the hairy edge of technology it's what happens. Welcome to the frontier!
posted by straw at 9:09 PM on April 22, 2011
The scaling thing is the most obvious issue here; fix that before anything else. supercres is right about all the other tips and measuring your impedance is a good idea. Get aggressive with that NuPrep :) But first and foremost set that Y axis the same on both computers. If it scales automatically, try to override it. Its a good idea to make sure your husband is not contaminating the signal with facial muscle movements, but I don't think high delta range activity like you're seeing is muscle; that's usually beta, much higher frequency. Make sure he's fully alert, well-rested, not bored.
It is weird how much the delta and theta range is overrepresented relative to the alpha range in his spectrum plots; that does make me think that there is a problem, but he could just have a weird natural power spectrum; they are heritable and some people have a peak offset a bit from 10hz.
About filtering; you may not have filtering built into that software, and in a lab setting we almost always filter our raw data after the fact with this expensive statistics and computing software called MATLAB. It comes with libraries of built in filters or you can write your own, but that's alot of math and work. Filtering is probably not something you can do with the kind of setup you're running, and for your purposes it probably doesn't matter; you just care about change after a session of neurofeedback, and as long as you can quantify that change and your noise isn't obscuring change from one session to the next, you're fine because you're not going to publish your data.
Is your husband sleep deprived at all?
posted by slow graffiti at 10:03 PM on April 22, 2011
It is weird how much the delta and theta range is overrepresented relative to the alpha range in his spectrum plots; that does make me think that there is a problem, but he could just have a weird natural power spectrum; they are heritable and some people have a peak offset a bit from 10hz.
About filtering; you may not have filtering built into that software, and in a lab setting we almost always filter our raw data after the fact with this expensive statistics and computing software called MATLAB. It comes with libraries of built in filters or you can write your own, but that's alot of math and work. Filtering is probably not something you can do with the kind of setup you're running, and for your purposes it probably doesn't matter; you just care about change after a session of neurofeedback, and as long as you can quantify that change and your noise isn't obscuring change from one session to the next, you're fine because you're not going to publish your data.
Is your husband sleep deprived at all?
posted by slow graffiti at 10:03 PM on April 22, 2011
Response by poster: My husband is 65 years old and quite sleep deprived!
posted by DMelanogaster at 8:09 AM on April 23, 2011
posted by DMelanogaster at 8:09 AM on April 23, 2011
Sleep deprivation alters the normal EEG in several ways and even chronic sleep restriction by missing as little as one hour of sleep a night over weeks causes problems. The shift toward the delta and theta frequencies may be partly explained by this. I still don't get why he has so little apparent alpha, but sleep disturbance definitely does weird things to EEG. If he gets more sleep you will likely see some kind of shift in the power spectrum.
posted by slow graffiti at 1:10 PM on April 23, 2011
posted by slow graffiti at 1:10 PM on April 23, 2011
Response by poster: Well at the point of that screenshot we were running around the room etc. trying to get a good signal, NOT an "alpha state."
posted by DMelanogaster at 3:50 PM on April 23, 2011
posted by DMelanogaster at 3:50 PM on April 23, 2011
This thread is closed to new comments.
posted by adipocere at 9:02 AM on April 22, 2011