Every hobbyist developer eventually attempts the same rite of passage: automating something that takes thirty seconds to do by hand. Mine was the cat feeder. This is the honest post-mortem — what I built, what broke, and what I would tell past-me before buying a single component.
The requirements (written by the stakeholder)
The stakeholder is a cat. Her requirements were strict: food at 8:00 and 18:00, portions measured to the gram, and zero tolerance for server downtime. My requirements were looser: learn a Raspberry Pi properly, avoid cloud subscriptions, and build something with a physical switch that makes a satisfying sound.
The build, version by version
v1 — the servo of lies. A cheap micro servo rotating a plastic cup. It worked eleven times, then the servo’s plastic gears stripped under the weight of actual kibble. Lesson: hobby servos are for demonstrating ideas, not for lifting things daily.
v2 — the auger. A rotating corkscrew design in a PVC pipe, driven by a stepper motor. This is how commercial feeders work, and now mine does too. The stepper never skips if you feed it from a proper power supply — the first attempt on USB power skipped steps and dispensed “portions” of pure chaos.
v3 — the boring version that works. Same hardware, plus a real food-grade container, an optical sensor to detect jams, and a cron job instead of my hand-rolled scheduler. The cat does not care that v3 removed my favorite clever code. Users never do.
What broke, in order of embarrassment
- Kibble physics. Irregular shapes bridge and jam. The fix was a vibrating mini-motor and a wider funnel — pure mechanical engineering, zero code.
- Time zones. The Pi trusted its clock, which was wrong after a power cut. Feeding time became a surprise. NTP and a tiny RTC battery module fixed it forever.
- The clever scheduler. I replaced it with cron. It has not failed since. My pride has recovered; my logs are quiet.
What it cost, really
Parts: about $45. My time: many evenings, which were the point. A commercial feeder costs $60 and works in ten minutes — and building my own taught me GPIO, systemd timers, power budgeting, and the profound wisdom of buying the food-grade container first.
Would I do it again?
Tomorrow. The feeder has run for months without intervention, the cat has forgiven the v1 starvation scare of 2025, and the whole thing publishes its status to a tiny dashboard on the hallway wall. If you are choosing a first hardware project, choose one with a stakeholder who cannot file bugs — just buy better motors than you think you need, and listen when the cat complains.
Read next: Debugging with a Cat on My Keyboard: What My Feline Coworker Taught Me About Patience