Thought I'd cross-post this to see if there's any interest:

Charles wrote:
>I was reading NUTS & VOLTS the other day, and seen an ad for
>the new basic stamps put out by MicroMint (the PicStic) and
>immediately thought that they would make an excellent vehicle
>for updating the HAL-4 and a construction project for Circuit
>Cellar Ink.  You could call it the HAL-9000!

I think the HAL used an 8031 to interface the PC running x86 fourier
analysis & display code to the A/D. So software would not be quite
compatable with the PIC Stump's basic.

>The reason that the PicStic is so nice is that it uses compiled
>basic, so it is capable of executing about 50k instructions a
>second, versus the 2k or 4k of the original Basic Stamp.  Plus
>they have one version that has a 2-channel, 12-bit ADC.

At what cost? Its a shame not to use the PIC bare with its indiginous A/D &
mux, going over 1 MIP. If a little assembly doesn't frighten you :)

>The architecture I propose would be capable of making an
>inexpensive (so to speak) up to 18 channel EEG (if GSR & heart
>rate were not implemented).
...
>The Communications Card would poll the other cards (via serial
>comm) and construct a master data packet that contains all the
>binary information for use by the PC.  With the analog data
>cards constructing sub-data packets from their ADC's.
...
>Of course this design is contingent upon being able to use all
>of the I/O lines of a basic stamp for serial communications (in
>a polled fashion).
...
>Hope you find this idea intriguing, and please feel free to use
>this in any way you see fit!

Sure do. I would suggest having analog interface and isolation cards plug
into a backplane. A PIC on the backplane samples and formats the data for
transmission to a PC. Maybe a special PIC output card or two to generate
audio or video for feedback. Yes, I've seen PIC's used as video generators
to interface a weather station anemometer to a television set. Cheap, dirty,
& clever.

In the same spirit, a PIC could generate signals that would operate analog
switches in a multiple switched capacitor filter. The signal products being
integrated & sampled form the EEG bands, which are serialized for a PC or
other feedback device.

Having a series of STAMP daughter boards may provide more channels, but
little reduction in complexity or cost. With data bandwidth so small
(<100Hz) why not just mux them?