This is the first log for this project. I've been writing progress updates on the journal at least once a week, and now that v3.0.0 is out, it's time to write it up properly here on Hackaday too.

Patternflow is an open-source LED synthesizer built from a 128×64 LED matrix, four rotary encoders, and an ESP32-S3. Think of it as a present-day take on Nam June Paik's Participation TV. Turn a knob and the creative-coding pattern running on the matrix shifts in real time under your fingertips.

I tagged v1.0.0 back in April, and I've rebuilt the thing from the ground up three times since. v3 took the longest by far. I didn't expect it to take this long. I figured fixing one problem would be the end of it. Instead, the more I fixed, the more there was left to fix.

Parts I didn't actually need

The biggest change is that I pulled every SMD part off the PCB. They'd been there in the first place because an AI assistant told me they were needed, so I just went with it. Then one unit broke mid-exhibition, and I remembered someone in the community mentioning that desoldering the encoder cap and resistor hadn't caused any issues for them. Nothing to lose, so I pulled them and tested. Worked fine. So I stripped it down to through-hole only. Cheaper to build, and more to the point, it's now hand-solderable start to finish, which matters a lot if this ever becomes a workshop kit. (More on this in Refocus and Experiment After Experiment.)

USB-C, and the one I fried

For the retail version I added a USB-C power adapter to the board. Turns out that's a nightmare to hand-solder. I actually shorted and burned one out. So the finished unit (the one going out through Crowd Supply) keeps the USB-C, while the DIY community version keeps the original two-pin screw terminal. While I was at it, I rearranged the module layout and shrank the board overall. (The full mess is in Worn Out.)

Printing the enclosure in one piece

The case used to be printed top and bottom separately and glued together. This time I redesigned it with snap-fits so a large-format printer could print it as a single piece. Sounds simple. In practice it was "revise, send it to the school's printer, wait a few days, check the failure, revise again," on repeat. That uncertain waiting was the hardest part of v3 by a wide margin. (I wrote about that loop in The Next Step.)

Crowd Supply is taking longer than I thought

Now that the hardware's done, let's talk funding. I signed the Crowd Supply contract a while back and figured it would move fast. It hasn't. To move past pre-launch I need 150 subscribers, and even after emailing the news directly to about 100 people already on the waitlist, only around 15% converted to a subscription. Honestly, that surprised me. These were people who'd already raised their hand once. Turns out crowdfunding is a much slower game than I expected.

Still, open source is starting to actually work

The numbers are slow, but there's a good sign underneath them. Lately other artists have started picking Patternflow up, reinterpreting it as their own project, and sharing that back out. It feels less like selling a device now and more like something actually spreading. That was the whole point of going open source in the first place, and it's starting to happen, bit by bit. (More on that in I Hope It Works Out.)

What's next

I think the hardware is solid enough now to carry the Crowd Supply launch. From here I'm shifting focus to software and community. Right now the community runs on Discord, and it turns out Discord is banned in more countries than I expected. So I'm rebuilding the main hub on the web instead, not killing Discord, just moving the center of gravity. Pattern sharing will probably be the core of it. (This week's thinking on that is in v3 and Beyond.)

The four-month hardware slog is over. Now it's time to think more and move less. I'll keep logging progress here and on the journal.