From: Jim Peters (jim_at_uazu.net)
Date: 2002-01-28 20:34:55
yaniv_vi wrote:
> what about testing vs a software simulator ?
> it could be easily built , and very usefull for testing .
> also you can test inside matlab or some tool like this .
> why you have to have h.w. ?
Because I want it, and because I've already fed sine-wave tones and
sweeps into my filters, and I've tried Jim Meissner's recorded data --
I've done software simulation, and now I need some real live data.
I think the best simulation of how a brain responds is a real brain,
so that is what I'm going to use. If you think you can simulate a
brain in software, please go ahead. I think it would be an immense
waste of time.
You must be completely crazy if you think someone can write this
software without ever seeing any live data and how it responds. Sure
I could write a huge program based on a load of theory that turns out
to be completely off the mark when it is connected up. Would you
consider writing a video game and never play-testing it until the
release date ?
Also, where is the interest and fun in writing a purely theoretical
program ? That's sterile to me. With real data there becomes a real
reason for coding every line.
You know, I think everyone should be a bit more willing to experiment.
Open-source traditionally works with many eyes and many testers.
We've got at least several pairs of eyes looking at the electronics,
but so far only one or two testers (the designers themselves).
I wonder whether the whole thing could be split into sub-projects with
well-defined interfaces, let's say:
- RS232 optically isolating link (RS232 in, RS232 out)
- uC board + software (2+ chan analogue in, RS232 out)
- Input stages (electrode connectors in, analogue out)
- Electrodes
If we're aiming for a kit, we'd need PCB patterns for each of these,
along with component lists, plus build instructions and basic setup
and testing instructions too. Someone has to do this work before we
have our hardware ready for your average electronics enthusiast.
If things are in separate parts, then if the first attempt at an input
stage doesn't quite work properly, but everything else does, then
that's the only bit that has to be replaced with a new version. Are
there enough electronics people on the list to work on this in this
way, or would it just be down to Joerg and Andreas in any case ?
Not knowing the electronics, I don't know if this approach would be
feasible, and obviously if Andreas can produce a finished production
version in one step, that would be great, but it does put a lot of
weight on Andreas and his simulation software, and it means that
everyone else can do almost nothing to help him.
I really don't know the answer -- maybe all modern electronics is
designed this way.
In any case, I'm going to have a go at building a simple lo-fi EEG
interface in order to make some progress with the software. That
seems like the best way forwards to me.
Jim
-- Jim Peters (_)/=\~/_(_) Uazú (_) /=\ ~/_ (_) jim@ (_) /=\ ~/_ (_) www. uazu.net (_) ____ /=\ ____ ~/_ ____ (_) uazu.net
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:37 BST