Maker Pro
Maker Pro

Non-integrating low pass filter

Samote

Dec 7, 2012
6
Joined
Dec 7, 2012
Messages
6
Hello,
I am relatively new at circuits, but a few friends of mine we want to, sometime in the future, work on an EEG. I have read that there is a danger of having out-of control voltage and essentially shocking the user of the EEG through his brain which is, obviously, very bad.

A solution that came to mind is to not actually have a reference voltage on the head, but rather some insulating blob of material somewhere on a distant circuit. This would mean that the actual voltage signal might have a huge DC offset relative to its approximately 3 microvolt amplitude. (Correct me if I'm wrong). Thus, if we were to feed this through a differentiator, it would give us a nice signal centered around 0V so that it can be amplified without producing monstrous voltage values.

(The graph on this wiki article is an example of what we are going for http://en.wikipedia.org/wiki/Event-related_potential)

My main issue however, is that the signal we are going for is a low frequency signal, whereas the noise is potentially a high frequency noise, so differentiating it would be a terrible idea. Thus, I want to know, is there a way of feeding a signal through a differentiator while having something act as a low-pass filter?

Also, would taking a constant but external reference voltage help at all in the issue at hand, that is, the risk of shock?

Also, do you have any other ideas of how to counter the problem of taking a low-amplitude low-frequency signal with an arbitrary DC offset?

Also, would it be possible to have the electrode itself modulate to something like a flashing LED, and connect that via some sort of fiber optics to a photodiode and pick up the signal from there?

Thanks in advance
The project is currently at its initial stages, and any radical changes to the methodology are welcome

---
Edit:
I intend to use an arduino board to somehow read the final signal. I have yet to purchase one and start fiddling with it, so I am not too sure how that is going to work.
 
Last edited:

Laplace

Apr 4, 2010
1,252
Joined
Apr 4, 2010
Messages
1,252
Medical equipment has stringent requirements to provide galvanic isolation from the patient into the kilovolt range for both powerline faults and electrostatic discharge. This is the first requirement, above all other considerations. EEG signals from any point on the scalp are just a few microvolts when referenced to another point on the head. Noise signals over a broad frequency range from DC upwards tend to be superimposed equally on the EEG point and the EEG reference point with respect to body ground.

What makes the most sense is to AC-couple the EEG signals with a nanofarad-range capacitor to a high-impedance low-noise differential instrumentation amplifier having high common mode rejection. Set a gain of 30 to 40 dB to boost the EEG signal out of the noise so you have something to work with. There will need to be some consideration for galvanic isolation of the power supply for this pre-amplifier, and a linear isolator for for each EEG signal. I'm not sure just where your differentiator idea would fit in here, but I don't believe it could be part of the pre-amp.
 

Samote

Dec 7, 2012
6
Joined
Dec 7, 2012
Messages
6
Set a gain of 30 to 40 dB to boost the EEG signal out of the noise so you have something to work with.

Sorry, I'm a little new at circuit design: How would an amplifier boost the signal out of the noise? Wouldn't it also amplify the noise, which, I assume is much greater than the desired signal, here?
 

Laplace

Apr 4, 2010
1,252
Joined
Apr 4, 2010
Messages
1,252
At the 3 uV signal level there will be a certain signal to noise ratio, whatever it is. If that 3 uV signal is amplified to 300 uV without adding additional noise, then the 300 uV signal has the same SNR. Now suppose there is some further signal processing that adds 5 uV of noise. That added noise would completely swamp the original 3 uV signal but it would barely affect the 300 uV signal. So the 3 uV level is not something you could work with but with 300 uV you do have something to work with. Just a fictional example to illustrate a point that keeping the SNR high is best done at the beginning of the signal processing chain. Spend your money there.
 
Top