Beta-0 is cancelled and we couldn't be happier
I was actually really happy that this prototype failed. Well, it didn't fail, but we aren't shipping it. It highlighted some really important design and manufacturing improvements. We've been frozen on this rough design because the CAD tool we started off with choked at 98% completion. It was the right thing to do to finish the board, learn what we can, and then move onto a different tool. Six months later, here we are.
We had a problem with our solder-screen where it was hard to register the board and align the screen properly. This resulted in the smallest components not soldering well at all. It's not that we can't fix this, or work on the temperature profile, but this was going to keep being a problem. It would probably take us 40 hours of building prototype after prototype by hand and the desktop PNP only makes sense after you've built a few. So, based on how hard it was to reliably build the board, we decided to ditch these boards and get working on a new design on the new CAD tool.
I'm in love with KiCad. It's an open source schematic and PCB tool that I'd been avoiding for years. Every bad thing they say about the software is true, there are some horrible usability issues, but! and this is a big but: Everything is open, which means I can build whatever workarounds I want. Every CAD tool has some serious usability problems, even the best. Knowing that I can crack the code open and modify the tool makes me much more assured that I can build the tool to match our workflow, instead of the other way around. You'll see some big surprises from us in 2017, but for now, know that the flexibility of the tool allowed us to iterate through a really big set of changes.
The biggest change was that we split up functionality into horizontal PCBs. I call it Horizontal Eurorack, because that's all it really is. But you'll see why this works so well in a bit.
We settled on a modified Eurorack format. Normally there is 7.5mm between the mounting hole and the edge. We lowered this to 3mm to cut down on PCB costs. There isn't much of a use for the board beyond the mounting holes, especially on a thin slice.
We created fractional slices of the board, settling on 1/2, 1/4, and 1/8. It's totally possible to create 3/4, 7/8, or whatever combination you want, but we're starting with this to see how it works and feels.
By creating a modular system, it decouples the functionality from a complex unified whole. It allows for modification, mutation, and deviation independent of the other sections. Although it makes the total PCB cost higher than a single board, the lower cost of scrap and the separating of functionality more than makes up for it.
Here are our first few modules:
We're looking forward to rapidly iterating on these modules and creating various configurations (KNOBS! KNOBS! KNOBS!). It may seem like a small setback, but by doing the redesign sooner, we've moved the project to where it needs to go to create the most variety and flexibility.
As always, head over here if you want to learn more.