EEG via soundcard??

[JBilderback]

Anybody know if anyone might be doing a HAL type project
using a PC soundcard ??

[Laurent Lessieux]

Hi,

I think i will try to do it soon. I have some basic knowledge on how to do a
pcb and soldering.
The problem now is to do a PCB. I have downloaded Eagle 4.01 and done my
first board but i would like to avoid doing it on a two faces board and uses
only one face but it seems almost impossible.

I have ordered the component but i am still waiting for some.

For the beginning i will limit myselft to two channels instead of four
because it's easier for the software, you just have to use the stereo input.

But i think that with a sound card like the SB Live 5.1, we should be able
to reach 7 channels.

The software should not really be a problem for me. It will work on Win2k
with perhaps a 3D part showing the localisation on the head and the waves in
real times.

If you are interested , please feel free to contact me.


Laurent LESSIEUX
Senior Software Engineer.

"JBilderback" <jiva@humboldt1.com> wrote in message
news:3ABA77C4.5AF0A5FD@humboldt1.com...

Anybody know if anyone might be doing a HAL type project
using a PC soundcard ??

[Lovass L]

Hello,

About using a soundcard for EEG:

The root of this idea is a possible pre-amplifier before the soundcard input
(microVolts to milliVolts).

The main problem is the wave-band.
A sound card's tipical low frequency limit is about 20 Hz. (Hi-Fi standard)
Even if the soundcard`s output is capable for low frequency effects (Dolby
Digital ,etc.), I doubt that even by multiplexing low freq channels input
can directly handle these freqs:

The EEG requires bandwith pass of 1-30 Hz, and the freqs from 1-14 Hz are
the most important.

And there are some minor problems as well e.g. 50 or 60 Hz electric 'noise'
filtering, galvanic separation for safety, etc.

Or do you have special ideas to bypass these problems?
(FM, freq multiplication, etc.)

Regards,

Lovass Laszlo

Hi,

I think i will try to do it soon. I have some basic knowledge on how to do
a
pcb and soldering.
The problem now is to do a PCB. I have downloaded Eagle 4.01 and done my
first board but i would like to avoid doing it on a two faces board and
uses
only one face but it seems almost impossible.

I have ordered the component but i am still waiting for some.

For the beginning i will limit myselft to two channels instead of four
because it's easier for the software, you just have to use the stereo
input.

But i think that with a sound card like the SB Live 5.1, we should be able
to reach 7 channels.

The software should not really be a problem for me. It will work on Win2k
with perhaps a 3D part showing the localisation on the head and the waves
in
real times.

If you are interested , please feel free to contact me.


Laurent LESSIEUX
Senior Software Engineer.

"JBilderback" <jiva@humboldt1.com> wrote in message
news:3ABA77C4.5AF0A5FD@humboldt1.com...
Anybody know if anyone might be doing a HAL type project
using a PC soundcard ??

[Laurent Lessieux]

Hi,

Thanks for the advice, and you are right about the low frequency limit, it
was hard to find but the SBLive have indeed a limit of 22Hz .

I have been investigating all this afternoon the problem and two simple
solutions are possible: AM or Frequency multiplication. I am not sure what
solution i will use but it seems that in both cases the software will do
most of the job anyway.

I am still looking at how to do both solution, but i think i will try first
to have the preamplifier done so i will have the signal (with the 50 Hz
filtered) and after i will try each solution.

Regards

Laurent.
>

[Mohan Prabhu]

Hi,

Lovass L is right. The sound card works from 20 Hz to 20 Khz.
The other concern of AC noise 50/60Hz is well pointed.

I am also looking for any circuits that will amplify very low freq 1-14Hz
and very low voltages (microV to mV) and filter out 50/60Hz.

In my opinion, the way to do it is to modulate a audio carrier and then
demodulate it in the PC.
I have attached a Windows oscilloscope that works with the sound card
if it helps.

Thanks
Mohan Prabhu


Lovass L wrote:

Hello,

About using a soundcard for EEG:

The root of this idea is a possible pre-amplifier before the soundcard input
(microVolts to milliVolts).

The main problem is the wave-band.
A sound card's tipical low frequency limit is about 20 Hz. (Hi-Fi standard)
Even if the soundcard`s output is capable for low frequency effects (Dolby
Digital ,etc.), I doubt that even by multiplexing low freq channels input
can directly handle these freqs:

The EEG requires bandwith pass of 1-30 Hz, and the freqs from 1-14 Hz are
the most important.

And there are some minor problems as well e.g. 50 or 60 Hz electric 'noise'
filtering, galvanic separation for safety, etc.

Or do you have special ideas to bypass these problems?
(FM, freq multiplication, etc.)

Regards,

Lovass Laszlo

Hi,

I think i will try to do it soon. I have some basic knowledge on how to do
a
pcb and soldering.
The problem now is to do a PCB. I have downloaded Eagle 4.01 and done my
first board but i would like to avoid doing it on a two faces board and
uses
only one face but it seems almost impossible.

I have ordered the component but i am still waiting for some.

For the beginning i will limit myselft to two channels instead of four
because it's easier for the software, you just have to use the stereo
input.

But i think that with a sound card like the SB Live 5.1, we should be able
to reach 7 channels.

The software should not really be a problem for me. It will work on Win2k
with perhaps a 3D part showing the localisation on the head and the waves
in
real times.

If you are interested , please feel free to contact me.


Laurent LESSIEUX
Senior Software Engineer.

"JBilderback" <jiva@humboldt1.com> wrote in message
news:3ABA77C4.5AF0A5FD@humboldt1.com...
Anybody know if anyone might be doing a HAL type project
using a PC soundcard ??

[Lovass L]

Hello,

I have consulted with my friend (HW expert) about the possible solutions of
soundcard EEG.

1. How it would be possible to get 7 channels? Most of the soundcards
practically have one stereo audio input (2 non-GND-symmetrical channels).
The MIC and the LINE IN inputs - AFAIK - are mixed by the sound mixer, so
it's very hard to count on them as separate inputs.

2. Galvanic separation (= using opto couplers for analogue sign, which is
not as easy as a pie is strongly recommended:

- for life protection, against the risk of possible electric shock
- elimination of electric noise from the computer

3. Practically the best (distortion free) band transfer method is the FM
(freq modulation), using one carrier per channel (for making 2-channel EEG).
Using multiple carriers (possibly for 4, 6.. channels) could make accurate
channel sign separation by realtime FFT very hard.

4. An additional idea is to use the analogue joystick inputs instead of
audio input, which makes possible to use 4 channels directly without
modulation! These inputs are adequately fast AD converters measuring
resistance of joystick pots. The holding time of the 4 bits are proportional
to the resistance value:

time (usec) = 24.2 + 0.011 * R {ohm}

In this case the pre-amplifier and the band filters have to be prepared with
the correct output impedance matching the range.

Conclusion: making soundcard-based EEG is not as simple, cheap and easy as
supposed. Even if joystick input (4) is used instead of audio band (3),
galvanic separation (2) is necessary. After all, the circuit and the
software could be more complicated than a multi-channel amplifier with some
FFT. And the capabilities of this EEG extension are rather limited.

Of course, these days it is possible to make a smaller, cheaper, simpler
low-consumption complete multi-channel EEG using serial (or infrared) port -
comparing to the nice but old design of HAL-4.

For regular use, the greatest problem is the annoying head electrode wiring.
Although I do not wish to be assimilated by the Borg from Star Trek or live
like Neo ), some integration ideas (circuitry and electrodes built into a
cap or helmet without wires) could be the most welcome.

Best regards,

Lovass Laszlo

[JBilderback]

1. How it would be possible to get 7 channels? Most of the soundcards
practically have one stereo audio input (2 non-GND-symmetrical channels).
The MIC and the LINE IN inputs - AFAIK - are mixed by the sound mixer, so
it's very hard to count on them as separate inputs.

it seems to me "only 2 channels" would be a worthwhile project if the
additional hardware required to get the project functional was
also very low cost. 2 channel machines are sold these days.

4. An additional idea is to use the analogue joystick inputs instead of
audio input, which makes possible to use 4 channels directly without
modulation! These inputs are adequately fast AD converters measuring
resistance of joystick pots. The holding time of the 4 bits are proportional
to the resistance value:

my limited experience with the game port (monitoring a thermistor) is that
the A>D is crude and unstable at best. I'm not sure how well they'd work for
this purpose.

[JBilderback]

I have been investigating all this afternoon the problem and two simple
solutions are possible: AM or Frequency multiplication. I am not sure what
solution i will use but it seems that in both cases the software will do
most of the job anyway.

I am still looking at how to do both solution, but i think i will try first
to have the preamplifier done so i will have the signal (with the 50 Hz
filtered) and after i will try each solution.

Joerg Hansmann in msg 118 @ buildcheapeeg@egroups.com

made the following comments: (which I don't understand)

---------------------------

(c input decoupling of the soundcard could cut off low frequencies.

A simple time multiplex encoding could be used, realized with a
counter
and a analog multiplexer. The multiplexer would scan through
some
reference voltages (e.g. AGND, VREF) and then through the
channel
outputs.
e.g.

; a sync signature:
AGND
VREF
AGND
VREF
; the eeg-channel data framed by reference voltages:
CH1
AGND
CH2
VREF
CH3
AGND
CH4
VREF
CH5
AGND
CH4
VREF

this pattern of 16 voltages would be repeated e.g. 256 times
per second.
However the output of the soundcard would have to be decoded
by software. The minimum sample-frequency for easy decoding
should be minimum 4 times the data rate(256Hz*16 Voltages),
that is 16384 Hz.

(d As far as I understand (most?) SCs have a minimum
adjustable sample frequency of about 8kHz,
so maybe one would have to take only samples of the
sample.

With the method mentioned above that would be no problem.

---------------------

Maybe you'd like to discuss this and your ideas with him?

[Laurent Lessieux]

Thanks for the information.

I think i understand what he means and the signal will be quite easy to
decode.
I have joined the group for the cheap eeg too.

The most difficult part for me is the electronic part, because i have a very
basic knowledge of this kind of stuff.
However the software part is quite easy for me.
I have been developing medical applications for the last 5 years in C++ for
Windows.

Making a eeg is probably not the easier way to learn

For the joystick port, i had also some very bad experiences with it. It
doesn't seem very reliable.

[Mohan Prabhu]

Hi,

I had e-mailed few people to find out using sound card to read
low freqs. of brainwave. Here is one of them what I have:

I received a reply from Dr.Steven at Cirrus Logic in U.K.

Quote:

Two suggestions:

Use a small Analog to Digital converter module which you can
plug into the printer port on a PC, which comes with some analysis
software.
I think PICO technology make one, see Electronics World magazine for
advertisements.

OR

Increase the value of the DC blocking capacitors in the input
circuitry in the
sound card. Usually the codec chip and drivers pass dc-20kHz, it is
the external
dc blocking capacitors that determine the 20Hz Hi Pass corner
frequency.
Making these capacitors larger will lower the corner frequency.
You still need to block DC, so the capacitors are necessary.

You will a high gain preamplifier to amplify the brain signals.
Also, think a lot about safety. Professional EEG units are
very carefully designed so there is no possibility of electrocution.
PC's are not designed to the degree of safety required by medical
equipment.

Good luck

Steve
Dr. Steven Harris, Director || Cirrus Logic Inc.
Email: snh@c... || Anglers Court
Tel Direct: +44 (0)7710 657828 || Spittal Street
Fax: +44 (0)870 0520252 || Marlow SL7 1DB
http://www.crystal.com || England



JBilderback <jiva@humboldt1.com> wrote in message
news:3AE83003.17921DE9@humboldt1.com...

1. How it would be possible to get 7 channels? Most of the soundcards
practically have one stereo audio input (2 non-GND-symmetrical
channels).
The MIC and the LINE IN inputs - AFAIK - are mixed by the sound mixer,
so
it's very hard to count on them as separate inputs.


it seems to me "only 2 channels" would be a worthwhile project if the
additional hardware required to get the project functional was
also very low cost. 2 channel machines are sold these days.

4. An additional idea is to use the analogue joystick inputs instead of
audio input, which makes possible to use 4 channels directly without
modulation! These inputs are adequately fast AD converters measuring
resistance of joystick pots. The holding time of the 4 bits are
proportional
to the resistance value:

my limited experience with the game port (monitoring a thermistor) is that
the A>D is crude and unstable at best. I'm not sure how well they'd work
for
this purpose.

[JBilderback]

2. Galvanic separation (= using opto couplers for analogue sign, which is
not as easy as a pie is strongly recommended:

- for life protection, against the risk of possible electric shock
- elimination of electric noise from the computer

here's a pre-amp design that someone dug up and posted to buildcheepeeg list:
http://www.cn.stir.ac.uk/~bp1/eegviewer/preamp/

[Thomas Watts]

Has anyone actually hooked a signal generator up to a variety of sound cards
and TESTED them at low frequency? maybe the attenuation isn't as great as
everyone thinks and could be compensated for in the software...

"Lovass L" <prochess@elender.hu> wrote in message
news:3ae34352@ccweb.ntplx.net...

Hello,

I have consulted with my friend (HW expert) about the possible solutions
of
soundcard EEG.

1. How it would be possible to get 7 channels? Most of the soundcards
practically have one stereo audio input (2 non-GND-symmetrical channels).
The MIC and the LINE IN inputs - AFAIK - are mixed by the sound mixer, so
it's very hard to count on them as separate inputs.

through use of multiplexing, to present the channels to the input in
sequence, either automatically or under external control from the PC

2. Galvanic separation (= using opto couplers for analogue sign, which is
not as easy as a pie is strongly recommended:

- for life protection, against the risk of possible electric shock
- elimination of electric noise from the computer

3. Practically the best (distortion free) band transfer method is the FM
(freq modulation), using one carrier per channel (for making 2-channel
EEG).
Using multiple carriers (possibly for 4, 6.. channels) could make accurate
channel sign separation by realtime FFT very hard.

4. An additional idea is to use the analogue joystick inputs instead of
audio input, which makes possible to use 4 channels directly without
modulation! These inputs are adequately fast AD converters measuring
resistance of joystick pots. The holding time of the 4 bits are
proportional
to the resistance value:

time (usec) = 24.2 + 0.011 * R {ohm}

In this case the pre-amplifier and the band filters have to be prepared
with
the correct output impedance matching the range.

Conclusion: making soundcard-based EEG is not as simple, cheap and easy as
supposed. Even if joystick input (4) is used instead of audio band (3),
galvanic separation (2) is necessary. After all, the circuit and the
software could be more complicated than a multi-channel amplifier with
some
FFT. And the capabilities of this EEG extension are rather limited.

Of course, these days it is possible to make a smaller, cheaper, simpler
low-consumption complete multi-channel EEG using serial (or infrared)
port -
comparing to the nice but old design of HAL-4.

For regular use, the greatest problem is the annoying head electrode
wiring.
Although I do not wish to be assimilated by the Borg from Star Trek or
live
like Neo ), some integration ideas (circuitry and electrodes built into
a
cap or helmet without wires) could be the most welcome.

Best regards,

Lovass Laszlo

[brianl703]

The Telediagnostic Systems T10 is an interesting device which I own a few of (having gotten them very cheaply off of Ebay).

It is a telephonic EEG transmitter. The oldest one I have is circa 1982, which should give you an idea of how not-very-sophisticated this device is.

It transmits 10 channels (8 eeg, 1 ecg, 1 eog) through a telephone line.

There is a receiver which decodes the 10 channels and is connected to whatever device is being used to view or write the EEG data.

As you may or may not know, a telephone line is sampled at a 64kilobit/second rate...8000 samples per second, with 8 bits per sample.

A telephone line's frequency response is normally specified as 300Hz-3000Hz.

This means that a sound card should be more than adequate for the purpose of getting EEG data into a PC...even for 10 channels...and the sample rate needn't be higher than 8khz.

Unfortunately, I had no luck getting any sort of documentation for this device from Telediagnostic Systems. They told me that it's "no longer supported".

I did disassemble one and it has an XR2206 for each channel which I believe is being used to FM modulate the EEG signal. Google XR2206 and you'll find that it's a "monolithic function generator" made by Exar. I remember seeing a schematic for a dumb Bell 103 compatible 300bps modem that used the XR2206 in the transmit circuit.

[Somnonaut]

the documentation for the T-10.
I work in a Sleep lab and we used to use this for in-pt studies and send it over phone to lab. Very time consuming.
As far as I know the modulation was of the multi-plexing of the signals, as there were fixed bandwith EEG, EOC (ELectrooculogram) EMG electromyogram, and other slow AC signals (respiratory) as well as 1 DC Oxygen Saturation signal in all of it.
Oh and also an EKG signal too.
I am not sure if I have the actual schematics, though I may, I am a friggin packrat.
What are you all looking to get out of the EEG?

[Somnonaut]

Here is link to OPEN EEG working on a soundcard.
http://openeeg.sourceforge.net/doc/hw/sceeg/