Re: [buildcheapeeg] Re: I published current design at SourceForge

From: Doug Sutherland (wearable_at_earthlink.net)
Date: 2001-12-03 11:06:40


Hi Yaniv,

> i'm looking for simple way for our group to offer a 1
> channel, very low cost, nfb machine, both pc based and
> stand alone.

I have to say I'm a bit confused by the different versions
that are on the OpenEEG site. There is the PDF describing
a multi-parameter biofeedback device based on TI parts.
Then there is the ComADC design based on 555 timers. Then
there is the ModularEEG design using Atmel microcontrollers
and fast IR. I also see something called RS232EEG, not
sure if that is the same as ComADC. So where are we going
with all of these designs?

Regarding your statement above, I wonder how much thought
has gone into target audience, who are these for, and what
will they be used for? One thing I think is important is
that the design doesn't go too low-end. For example not
using a microcontroller introduces a lot of issues but
microcontrollers are not expensive. The minimal performace
specs might be different for someone trying to do alpha
training vs ADD or addiction treatment vs just something
neat to play with, or for entertainment.

I like the idea of a vert cheap and also small and low
power EEG, even for my own use. I want to try some
experiments in long-term monitoring with me as the
guinea pig, and I need a small/simple/low power EEG to
do that. So I'm definitely interested in the low cost
simple design. However, I caution not to go too low
end, or it becomes a toy. Let's not make another HAL-4
neat toy EEG here. HAL-4 has a frequency response that
pretty much misses the delta and beta spectrums, and
it has a poor sampling rate.

> it's very important that the work that need to be
> done in order to have that would be minimal , and
> won't require expensive development tools, due to
> the nature of volunteer groups.

I'm all for that. The Atmel microcontroller tools are
free (I think), and on the PC side I am really keen
on working on Linux apps (GNU). However it's clear
that windows versions are needed too, I think there
are some free borland C tools kicking around.

> so on thing to do is to start from what we have -
> joerg's comadc, and go from there.

We need a microcontroller between that and the PC.
That design requires disabling interrupts. Joerg
does that by running in DOS only mode, effectively
taking over the whole machine. I don't know many
people who want to go back to DOS. We need to remove
that limitation or we are too low-end in my opinion.

Here's what I want to understand: what are the
specs on band width (frequency range) and sampling
rates for these different designs? So far nobody has
answered those questions. How does ComADC compare to
ModularEEG in terms of band width and sampling? Is
there perhaps a possible target in between the two?
ModularEEG has lots of parts while ComADC has very
few. This is just my opinion, but I think that the
real target might be somewhere in between. The amp
section for ModularEEG looks good, I would rather
see a simplified version of that that ComADC.

> another idea to make it much easier is that at first
> step, it would be only pc based and be working good,
> and very cheap, but anybody who have pc,

I don't see any standalone features in any of these
designs yet. What would be the method of feedback?
Audio? Graphs on a small LCD? Perhaps there could be
two boards that plug together, one is the EEG that
plugs into PC, the other board is a feedback board
for stand-alone operation, one simple ribbon cable/
connector joins them.

> could get from web the right development tools,

Yes, I think it's important to avoid commercial
devtools that cost, and I also think that you are
much more likely to succeed with an open source
type license. Anything less will discorage people
from contributing. Look at the success of linux
as your example.

> and could develop software for the standalone
> version and download it via pc. so in such a way,
> the work could be shared by large group of
> volunteers , in order to have the standalone
> version. again empasize on cost, and minimal
> work is very critical.

Yaniv: what is the business model? Who gets the
money and where does it go? This needs to be made
clear before many people will dive in. I am not
interested in money from this, but I do want to
understand the business model if I am planning
to contribute development time.

> anybody would like to help in thinking and /or
> designing / coding for such a thing?

I am quite experienced in software, including
embedded and microcontroller plus PC stuff, and
I've done quite a bit of research on EEG. I am
willing to help with the software, but there are
the big issues for me:

1) Which design is it that we are working on?
2) Is the hardware in alpha, beta, or what?
3) We need hardware to work on this stuff, so
4) How can we get PCB boards made, and
5) How do I get the firmware?

If I can get a blank PCB with the target design I can populate
it myself with the parts. If the firmware is available I can
easily flash the uC and start programming for the PC. Jim has
some source code for his HC11 system, I think we can leverage
some of that if he's willing to share it. But we really need
to get hardware into the developer's hands before you are
going to see much software materialize. I suggest that each
throws in a few bucks and we do a low volume order of the
blank PCB board from PCBexpress or a similar place. I don't
see how you are going to get very far without producing some
PCB boards, we will just keep talking forever. Can someone
answer the above five questions?

-- Doug



This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:32 BST