Re: [Openeeg-list] first comparisons of active/passive electrodes

[Jerry J. <Jerry@JerryJoplin.com>]

Jarek wrote:
> It is not only matter of processor speed, but also memory. Besides avr is already quite
> busy with sampling and sending/receiving data. And it is crucial, that the sample timing
> is not disrupted. 

I'm betting the AVR is in sleep mode for the vast majority of the time. 
The PWM and UART only interrupt the AVR briefly, even with 6 data 
channels and 256 Hz sampling I bet it is sleeping over 90% of the time. 
  In fact, it seems the sampling rate could probably be cranked up quite 
a bit if it weren't limited by the UART transmission rate.

However, I think you're right about the memory constraint.  I've been 
working on a proof of concept on an AtMega163 to perform integer FFTs, 
and it only has 1k of available SRAM, which limits the sampling size 
down to 64 sixteen bit samples.  The AtMega128 has 4k, which would help 
quite a bit.

> Something like that would make sense, if we did not send data to PC, and AVR was
> responsible for everything. But if PC is involved, then it is better (IMO) to have as
> much logic there as possible. It is simply much more flexible.

I guess I'm thinking about moving as much logic into the device as 
possible in order to make a plug-n-play USB based HID device.  This 
would offer the greatest amount of flexibility in integrating the EEG 
data into any application.  But again, I'm not sure if this is going to 
work as I just got started on openEEG.

Jerry