Close

Scaling, coding, etc..

A project log for NeuroBytes

Build your own nervous system!

zakqwyzakqwy 03/04/2016 at 21:590 Comments

Progress! I ordered the stuff for a 210-piece NeuroBytes v0.8 run. They'll be light blue with white silkscreen, just like the v0.4 boards from 2014; however, they'll ditch the bumpy silver HASL coating for sleek ENIG gold. Minimal changes from the six prototype boards beyond the color; I futzed about with the silkscreen a bit and pulled a lone 10k resistor into one of the 4x networks.

[Above: I had the PCB vendor take care of panelizing the v08 outline; it's routed with four mousebites per board. Not a completely optimal use of space, but having them take care of it saved me a good bit of time. Boards are scheduled to arrive in Minneapolis next Tuesday, 3/8/2016.]

The Digi-Key box showed up yesterday. While the 210-piece run will nominally require 1470 JST GH headers, quantity discounts meant it made sense for me to get my hands on my first full reel:

[also shown: 1000' spool of 26 AWG wire.]

I'm planning to build up a bunch of cables between now and Tuesday, and then get a few dozen boards churned out midweek. And yes, I did get a stencil this time (from OSH Stencils)--it's Kapton (captain?) so I'm guessing it won't last the whole run, but it'll do for now:

[in-focus stencil with blurry PCB rear layer in the background...]

I've also been making progress on firmware. The Izhikevich model works well but isn't ideal for demonstrating simple integrate-and-fire behavior between a few NeuroBytes, so I've been porting a copy of the v0.4 firmware over to the ATtiny88 platform. I've started using the 'swing the camera' debug method to track down pesky PWM glitches:

[the glitch is the single green flash between red flashes..]

Also, this photography method provides a great visual for bragging a bit about the firmware efficiency improvements I've implemented. I mean, let's be honest--the v0.4 firmware was _terrible_, and the new code could probably still be improved a lot, but the >4x refresh rate improvement is great:[taken at 1/13 s shutter speed]

So yeah. Work is cut out for the next week. Finish firmware, build cables, assemble boards, update GitHub repo. With any luck, we'll have a box out to our collaborators by mid- to late-week, and will have enough devices left over to show off at Maker Faire NoVa next weekend.

Discussions