Maybe there shouldn't be just _one_ ModEEGV2. Perhaps there should be
different versions for different purposes. I imagine the priorities are very
different for those working on neurofeedback (for meditation purposes)
devices compared to those working on BCIs.
Here is my personal list of the top two priorities for
1. Number of channels: I'd say 16 minimum. So a monopolar design would be
the most practical.
2. Ease of construction: Many researchers aren't going to have great
soldering / construction skills. Try to make it "plug & play" as much as
possible. This may require standardizing to one type of electrode.
Everything else would be lower priority from my perspective. I wouldn't try
to reduce the cost, just keep it reasonable.
Just my two cents.
From: Stefan Jung [mailto:bio.sens.eng@...]
Sent: Tuesday, January 26, 2010 7:18 AM
Subject: [Openeeg-list] ModularEEGV2
I am not your man to help on your project, I have only thrown some
ideas around. Whatever I do in life always has commercial undertones.
Development costs money and it ain't much, but I do things that I
like/enjoy with the possibility of perhaps, just perhaps making some
money. It does not have to be so much money that I can replace my
income. After all, like Lisa Minelli sings - money makes the world go
round, the world go round, ... For that reason I will not get involved
in a truly open project, I am willing to give, but not to such an
extend. I just don't have that much time.
I was not a member of openEEG when it went through the dev stages, but
it was a team consisting of many people. To drive a design like that
is a long winded exercise, you need the buy-in from team members. I
am my own little drummer boy, it would frustrate me to develop
something like that in such long winded fashion - I have enough of
that sort of thing at work (I work for a mil co. mainly involved in
missiles). EXG-2 was done 100% by myself, hardware design, firmware,
PCB design, the whole tut, and it was reasonably fast, I made
decisions and implemented them, a one-man-band. I enjoy that,
especially in contrast to my day-job.
Electronics is a hard business to make some money. The man on the
street cannot understand why your electronic gadget costs $300, when a
calculator costs $7 and even a fancy PVR only costs $200. The way
electronic gadgets prices fell over the last 10 to 20 years is totally
mind blowing. Sometimes one wonders how anybody can still make money.
But there ARE opportunities, they are just not that many.
For the above reasons I have scaled down my personal (non-work
related) electronic design activities and put my time & money in
property development, something with pritty much assured returns in
the long run.
I will start fading away from this list again, and now is a pretty
good a time as I only have so much time to go around. But here are
one or two more comments on the subject matter:
1. I had a look at the AT91SAM family at the time I started off with
EXG2. The chip itself is cheap but what I didn't like are the many
decoupling caps and voltage regulator(s) you need to add, just to get
the core running. The PIC18F needs one crystal and 4 caps and will
then merrily act as a USB device. The SAM is also not an "ordinary"
uP, I suspect it's learning curve will be a bit steeper. It certainly
is powerful and fast. Everything goes ARM, cellphones often use more
than one ARM processor... so it's also not a bad idea to get exposure
of this family - something to put on your CV :-).
2. I would consider using the QDS protocol which gives a payload of 8
channels at 16-bit resolution. It is supported by BE and you can get
the protocol in the exg.c module of my EXG2 source code distribution.
Later you can implement something else - if you find it, or if you can
convince someone to support your own home-grown protocol.
3. I don't know exactly what switching you have in mind. You could
add CMOS switches, they will possibly add some noise (all silicon
usually does). Any switching will deviate from the KISS objective.
QDS switches the low level signals with their "QDS 4 data switch" -
but I don't know if they use relays or CMOS switches.
4. A big hassle is having to short unused inputs. This is only
required if these unused channels partake in the DRL circuit. Also,
if you use active electrodes you can terminate your inputs with say
1Mohm in which case no short circuiting will be necessary.
5. Bipolar designs should really be kept to a very small number of
channels. If you want configuration flexibility you go monopolar.
With a monopolar design you can then add, subtract and otherwise
configure channels in any fashion you choose (in the digital domain).
You need to create a priority list first. You need to tell this forum
exactly why a new design. Then people can decide if they buy into
your vision and avail themselves to helping, or if they have a
different vision and prefer to be silent observers.
Why a new design? Is it to:
1. reduce the cost
2. make it easier to build
3. increase the signal-to-noise
4. increase the overload margin/headroom (i.e more bits)
5. be able to use it for other signals such as EMG/ECG/EOG
6. implement many more channels (16, 19 or 24 channels maybe?)
7. make it portable (i.e. wireless & battery operated)
8. have built-in impedance measurement
9. use build-in active electrodes (perhaps use only active electrodes?)
10. use it as a stand-alone device for BCI, gaming, etc. (thus needing
lots of processing power)
1. Low cost versus ultimate performance?
2. Modular versus high(er) performance versus cost versus compact design?
You will definitely polarize this community into opposing camps of
what each person thinks (and expects) openEEGV2 should or should not
The Planet: dedicated and managed hosting, cloud storage, colocation
Stay online with enterprise data centers and the best network in the
Choose flexible plans and management services without long-term contracts
Personal 24x7 support from experience hosting pros just a phone call away.
Openeeg-list mailing list
Go to the above address to change your
subscription options, e.g unsubscribe.