From: Joerg Hansmann (info_at_jhansmann.de)
Date: 2001-12-14 14:31:07
Hi sleeper75se,
----- Original Message -----
From: sleeper75se <sleeper75se_at_yahoo.se>
To: <buildcheapeeg_at_yahoogroups.com>
Sent: Friday, December 14, 2001 2:09 AM
Subject: [buildcheapeeg] Re: Necessity of ir isolation
> Hello all,
>
> --- In buildcheapeeg_at_yahoogroups.com, Moritz von Buttlar <info_at_baltic-microsolutions.de> wrote:
>
> > >I'm working on it... :-)
> >
> > Great ! Feel free to make the changes already in the schematics and
> the layout.
> > (Also all your other layout improvements, if you want).
>
> I'll try... I don't get along with Eagle very well though. It is very
> different from Protel, that I use every day.
> Just a thought: we can save some space if we replace the resistors
> and decoupling capacitors with surface mount parts. Size 1206 is
> pretty big (5mm x 1.5mm) so they should be fairly easy to solder,
> even for inexperienced people. What do you think?
>
> > I think Joerg knows some people in the department for medical
> > engineering / Luebeck University. They might even test our device
> > for us for free. Maybe they have access to the norm and we could
> > get a copy ?
>
> That would be perfect! Any knowledge and information they have is
> valuable.
>
> > Something else: What about the input stage of the amplifiers ?
> > Should we change it ? How can we change it to keep it safe ?
>
> Yes, I believe they need some work. The highpass filter reduces the
> input impedance to 10Mohm, and thats not very good.
That is not quite correct.
The difference-mode impedance is really reduced to 10Mohm - however this
type of impedance will only result in a minor amplification error
(about 0.1% loss) due to a virtual voltage divider between the
electrode impdance of ca. 10k and the diff-mode input impdance of 10M.
What is much more important for rejection of line hum and other common mode
signals is the common mode impedance - and this is the full INA114
10^10 Ohm input impedance. The seemingly asymmetric RC-network I have put
before the INA114 inputs has no diminishing effect on the common mode
impdance. At least in a MICROSIM (a pspice clone) simulation it works
as intended.
> Compare with the
> Brainmaster and its reworked input stage (10Gohm).
Do you have a URL ?
I only know the old bm input stage.
> I suggest that we
> lower the gain of the instrumentation amps to 10, put the filter
> behind them and then amplify the signal by 1000. The downside is more
> noise ...
This is the old design of RS232EEG...
and I have intended to build something better in the modularEEG.
> Safety: INA114 has an input impedance of 1Tohm, so a single 470k
> resistor on each input shouldn't do much of a difference for the
> measurements, but they will limit the current to safe levels if the
> battery gets shorted to the inputs somehow.
IMO this is a bad idea. It will introduce 18uV p-p noise due to
the drop of INA114 18pA p-p input noise current (for simplification
over a frequency range from 0.1 to 10Hz) over the two 470k protection
resistors.
(see picture in the attachment with noise calculations for different
instrumentation amps and different source impedances)
BTW. (random) noise is comprised (at least) of the geometric addition of following components:
1)Johnson noise in the electrode resistance (and any resistors in series)
2)source impedance * input noise current of the amplifier
3)input noise voltage of the amplifier
Regards,
Joerg
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:33 BST