May: Heads Down, and Picking a Show

Posted by KA9NWM

May was the month I stopped thinking about Helioclock as a side project and started building it like a product with a ship date.

That framing actually started a little before May 1st. Hamvention in Dayton runs every year in mid-May, for 2026 it landed May 15-17, and around the third week of April I did the math and realized there was no honest way to have a booth-ready product in three and a half weeks. I could have shown up with something half-finished, but Hamvention is the biggest gathering in the hobby, and a rough demo table there does more damage to a new name than good. So I made the call to skip it as an exhibitor and put the time into the product instead: press outreach to QST, CQ, QRZ.com, and eHam.net instead of a booth, timed to catch the HamClock end-of-life conversation while it was still fresh in the community’s mind.

That decision is what turned May into a pure build month.

By mid-May the SKU plan had firmed up into the shape it’s basically still in today: a preloaded SD card for people who already own a Pi 4 or 5, a USB installer stick for people bringing their own x86 PC, and a turnkey N100 appliance for people who want something that just works out of the box. Each of those needed its own installer script, and May is when install.sh and install-pios.sh both got real, iterative attention: first-boot flow, hostname setup, mDNS, the systemd services, the whole chain that turns a blank OS install into a running Helioclock unit with zero manual steps.

A few real decisions got made along the way, not just features added. The slideshow/auto-cycle mode, one of the very first Phase 2 features from back in April, got pulled out entirely on May 28th after I traced a slow memory leak back to it (repeated toDataURL calls building up over hours of uptime). It wasn’t a small feature, but it wasn’t worth shipping something that would eventually crash a kiosk running unattended for days. Around the same time, a stubborn news-feed interval bug turned out to be trapped inside the wildfire layer’s toggle logic, a good reminder that in a single-page app, features that seem unrelated can end up silently wired to each other.

May 28th was also the day the first real alpha unit, a Pi-based device I’d been calling N97 internally, shipped out to my first ready to run system to my long-time friend and alpha tester, Stuart WB6RXG. We tested running the update pipeline end-to-end over the air rather than a manual flash. WiFi detection, first-run setup, remote updates: all of it had to work on hardware that wasn’t sitting on my own desk anymore. That’s a different kind of nervous than testing on my own machines.

The month closed out with working satellite and radar, new map layers, Sea Surface Temperature and Sea Surface Temperature Anomaly, built from NOAA’s Coral Reef Watch imagery, which meant real server-side image processing for the first time: cropping out NOAA’s own chrome, cutting out land pixels, and re-centering a Pacific-projected image onto an Atlantic-centered map to match everything else in the app.

And right at the end of May, with Hamvention behind me for this year, I started looking hard at where I actually would exhibit. Huntsville Hamfest, the ARRL National Convention, August 22-23 at the Von Braun Center, was the answer. It gave me a real runway to be booth-ready instead of scrambling, and it’s an ARRL National Convention, which carries its own weight in this community. I sent the first exhibit inquiry that week. Everything since has been building toward that date.

73,
Anthony, KA9NWM

Leave a Comment