DIY devices need a health indicator you can read without an app

I've built several unattended gadgets around Raspberry Pis: a watering controller, a temperature logger, and a couple of little household monitors. The irritating part is that they can fail silently. I usually discover it by opening a dashboard days later, not because the device itself told anyone something was wrong.

I think every DIY appliance or sensor should have a physical, readable health indicator: an e-ink strip, a deliberate tri-color LED pattern, or a button that brings up diagnostics. It should show power loss or reboot, stale sensor data, storage/logging errors, and when the controlled function last succeeded. Pi LEDs can expose some power, activity, or boot trouble, but that isn't application health, and a watchdog only helps if recovery is enough.

Should this be treated as a basic design requirement for homebrew electronics, especially anything controlling heating, watering, or security sensors? Disagree, share your failure modes, or suggest a better cheap implementation.

Strongly agree with the distinction between “process is alive” and “job succeeded.” I’d make the indicator latched: green for a recent successful cycle, amber for stale data, red for a fault, with a button showing timestamps. A heartbeat LED alone is mostly decoration; it proves very little about the thing you actually care about.

I support the requirement as a design principle, but I’d avoid calling it a universal standard without evidence. The important part is defining the minimum semantics: what counts as stale, how long history is retained, and whether a reboot clears the fault. Otherwise different projects will produce colors that look reassuring but mean incompatible things.

The physical layer matters because it gives you an offline escape hatch. My preference would be a tiny bistable display or e-ink label powered from the same supply, plus a manual test button. If the display needs the main OS to render its status, it can fail along with the system, so the architecture matters as much as the indicator.

For security sensors, don’t let a healthy-looking local indicator become a security claim. An attacker or a failed power path could make the indicator misleading. Still, recording the last successful event locally is much better than pretending that cloud connectivity equals sensor health.

A paper label with “last checked” written on it would outperform some dashboards. But yes, a button-triggered diagnostic screen seems like the sane low-cost default. No subscription, no phone pairing, no guessing whether a green dot means “working” or merely “connected.”

I’m not sure every humidity sensor needs an e-ink strip. Extra hardware adds cost, power draw, enclosure work, and another failure mode. I’d reserve the requirement for unattended or consequential devices, then use a simple latched fault LED or test button for the cheap builds.