NeuroBytes v0.5 is a prototype in the extreme sense of the word. We wanted to test a number of new ideas all at once and lacked the patience for a custom PCB, instead reverting to a hand-carved copper board and many blue jumper wires.
A few areas we changed compared to previous models:
- Indication. Specifically, moving away from a single RGB LED for everything. Instead, we wanted to try using a series of tiny green elements arranged in a circle to represent the current membrane potential; a high-brightness white LED in the center of the circle for firing; and red/blue LEDs on each dendrite to show excitation and inhibition.
- Connectors. The v0.4 models had many issues: they were bulky; they were insertion mount devices unsuitable for future pick-and-place automation; they had locking tabs which we had to remove for usability; and they only had three conductors, which limited our inter-neuron communication flexibility. v0.5 used JST's SH model, a 1.0mm pitch SMD connector.
- Microprocessor. The ATtiny44A was nearly maxed out with v0.4, and the added indication requirements meant a higher I/O count device was necessary. v0.5 moved to the ATtiny88 in a 32-pin quad flat pack configuration, giving us plenty of pins for directly driving LEDs.
- Size. v0.4 used 0603 passive components; in an effort to reduce PCB expense and wasted space, we wanted to get a feel for 0402 stuff, including the green LEDs.
The finished prototype never really became a NeuroBytes board, mostly because we never took the time to completely wire the connectors or port the v0.4 software to the device. Furthermore, the lack of multiple prototypes meant the board wouldn't have any other nodes with which to connect.
Having said that, we still managed to learn a lot from v0.5--in particular, that multi-LED indication looked great, as shown in this mockup: