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

From: Doug Sutherland (wearable_at_earthlink.net)
Date: 2001-12-05 13:48:32


Hi Laura,

> It looks like the most difficult part of this is amplifier
> design. If I can't get specs on the current amplifier
> design and some sort of confidence that it is well tested,
> there is no way I want to get a custom PCB made to build

Well, I guess we just have to wait, or try one of the other
circuits. It sounds like the RS232EEG was tested and worked
well. Brainmaster design is definitely tested. Then there's
the simple Biosemi design, and also Jim's design.

> IMHO, if we are targeting build-it-yourselfers we need a
> design with less parts.

There seems to be a huge difference in parts count and
complexity between the ComADC and the microcontroller
based designs. It's like volkswagons and caddilacs.
Maybe you are right, where is the honda accord?

> The low-partcount biosemi design seems to have some
> advantages over the Brainmaster design (reference post
> 1264 for details).

I looked at that post and discovered that I wrote it!
Well, yes, it's pretty impressive that the biosemi
amp design yields better band width and CMMR than the
brainmaster freeware build specs, while using so few
parts. I discovered that the commercial brainmaster
had better specs than the freeware design.

> I'm going to run a cost analysis on parts for these
> two designs either tonight or tomorrow.

I'd like to see your parts list and info on cost and
suppliers if you do this. I'm really itchy to start
playing, so I might try the biosemi design just to
give me something to play with.

> The job of the microcontroller is not really very
> hard (for a person with embedded programming
> experience ;) Just sample the A/D and transmit the
> results to the serial port.

Yes, the hard part is the PC software. Raw data is not
going to be useful. Quite a bit of complex software
is needed for really useful feedback and training
protocols.

> it would be nice if we could support a couple of
> different ones including the Atmel AVR and the 68HC11.

I agree. But then the specs are all there, so I guess
you are free to grab an amp/filter spec and bolt it
onto your own microcontroller. I like what you said
about the modular approach though: it would be nice
to have the amp section complete on one board rather
than spanning onto the processor board.

> At some point we need to evaluate the current serial
> spec (see post 33) and decide if it is what we want...

It sure would be nice if there were standards such
that software could be used with different EEGs. But
then the standards wars are how people protect their
IP and markets, so it's probably never going to
happen. That's why microsoft has their own standard
for everything, they want you to be stuck in their
world forver.

> but right now, any EEG is better than no EEG!

Yeah. Well, I've played with a commercial Brainmaster
and I was mucking around with software for that. It
used to have a hacker community, and the specs and
firmware and PC code were freely available. But then
it turned into a more closed environment, you can't
even be on the discussion list without paying. So I
got a but turned off by that experience. I sold my
brainmaster and now I am once again itching to do
some measuring and software. Sounds like the new
amp design isn't quite validated and ready yet, so
I'm a bit torn. I could do one of the following
to give me something to play with:

1) Follow Tom Collura's brainmaster design
2) Try out the simple Biosemi design
3) Try out Jim's HAL-based design

I'm interested in other bio measurements too, so
I think I am going to order a bunch of different
instrumentation and op-amps and start playing
around. Maybe I should work on ECG while waiting
for the new ModularEEG to harden. But then Yaniv
said we should focus on ComADC so that leaves me
wondering ...

-- Doug



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