We've been rockin' a 4-week hardware+firmware iteration cycle which has been hectic to say the least. Talking to a lot of potential customers and end users, quickly integrating their suggestions into new models, trying new components, etc... it's been fun. Unfortunate side effect: my documentation efforts are woefully out of date.

NeuroBytes v0.6 and v0.7 have come and gone. If you want to learn more about either generation, head over to the section of our website that details prototype iterations and click through to the associated entries at the bottom. I'm going to skip recreating that information here as it's redundant at this point and I'd rather talk about the latest version!

[v0.8 final board design, 30mm x 55mm including protrusions.]

Notes:

• Similar electrical specs to v0.7: ATtiny88 in a VQFN28 package, five dendrites (inputs), two axons (outputs), single RGB LED. Adds back in the mode switch from v0.6. Pogo pads for programming with an associated (and yet-to-be-constructed) jig.
• Board outline is based on our logo, with the five dendrites extending out from the cell nucleus at "organic" lengths and angles. Top of the board is super clean--all the user sees is the neuron shape rendered in ENIG, while the NEURO TINKER name is soldermask-covered copper.
• The center of the gear on the logo is a 3mm hole, straddled on the back of the board by a bottom-mount RGB LED.
• Connectors are surface mount, right-angle, 1.25mm spacing JST GH. They have locking tabs that are easy to spot, and pop in and out easily once the lever is pushed. More details on how I build the cables in another post, once I have more raw materials in hand (tomorrow).
• No more massive hole on the back of the board for easier VQFN ground pad soldering; for these prototypes, I reflowed these chips in my toaster oven, touched up the edges with my trusty WES-51, and hand-soldered the rest of the parts. Worked well enough, although I'm still having trouble getting one of the boards to listen to commands. So call it 5/6 yield.

[#Stickvise with the crucial logo placement... well played, guys.]

We wanted this version to look more like a neuron; our logo was a good starting point, so we ended up using it as the basis for the board design. It means we can't v-score the boards to panelize them, but everyone we've spoken to agrees that the product needs to look like its biological analog. We're also serious about avoiding an enclosure--that adds cost and waste and reduces our flexibility in terms of future iterations (tooling is expensive!); hence, bottom-mounting all of the parts so the product is uncluttered while in use.

[please ignore the wonky filter cap between the LED and the IC...]

As you may have noticed in previous project logs (or on our website), we're working with Anj Petto at UW-M to trial the product in a freshman-level A&P course in April. So... at least for this round of prototypes, we also included a mascot to the left of the printed instructions. Helped us push the limits of OSHpark's silkscreen, and it ended up turning out quite well!

[full brightness green-channel test looks amazing!]

Up next: finishing the pogo rig for speedy programming, tweaking the design a tiny bit (specifically, adjusting routing radii to avoid undershoot), porting over the v0.7 firmware with a few bug fixes, updating the GitHub repo (finally.. ), and ordering a larger run of boards--this time I'll get a stencil, too. Stay tuned!