[Openeeg-list] Re: modular eeg work report

[Andreas R. <and...@st...>]

Hi Chris,

I read a bit about interference ...

http://www.biosemi.com/publications/artikel3.htm

First of all, the 50Hz interference is lower when there is no isolation to power ground - which explains why it starts working when you touch GND1, but not when you touch analog ground. However, for safety reasons, we can't use that method... :-p

Second, shielding the cables is very important - they pick up a lot of interference - so try fixing that first. You may have to use the metal box again, too.

Third, if the DRL is not working (this is really Jörgs territory - have you mailed him about it? He's a bit busy so he might not reply right away) you must connect an electrode to (virtual) ground (VREF/2) instead, or else you won't have any ground reference whatsoever.

About that 100uF capacitor: forget I mentioned it - it probably won't do much good. At first I thought your problem was related to high impedance in the battery, but now I think it has more to do with shielding...

A couple of remarks about the design:

Guarding was excluded in an effort to keep the design simple. There are EEG circuits that work without guarding, so we assumed it should work for us as well, particularily with the DRL. If we can't get your board to work correctly, we may have to rethink.... (A guard driver can be patched into your board so don't worry that you have to rebuild the whole analog PCB - I'll draw a patch-board for you.)

About the second 7805 in the new version of the digital board: The idea is to replace the DCDC converter with two batteries, for those who want to, and save the expense of the DCDC converter ($10). That is why the second 7805 and the DCDC converter should not be used at the same time. Since you already have a DCDC converter, you can ignore this completely.

> does this mean to zero the 50Hz-bin in the fft and retransform to time-scale
> ? could such signal processing be done 'on-the-run' at the avr  ?I'm heaviliy

Well, I was thinking about a IIR-filter (compare with FIR...) that runs on the PC and operates in the time-domain, not frequency domain.  However, in this case, it looks like the amplifiers are saturated with the 50Hz signal. No amount of digital filtering can get rid of that.

So for starters, try this, in order

a) Switch to shielded cables.
b) Add metal shielding for the whole amplifier board.
c) Disconnect the DRL electrode from the DRL-amplifier and connect it to virtual ground instead. Perhaps you can use a switch (make sure it is shielded) so that you can change between DRL and GND easily. That way it becomes easy to determine wheter the DRL works or not.

I really hope this helps... :-)
If not, there are at least two other people here to help you; Jörg and Jim Meissner (though neither has subscribed to this list yet.)

Regards,

Andreas