Ten Smoke Detectors, One Timer

Background

I've got ten of these things — First Alert ZCOMBO, Z-Wave smoke and CO combo units, installed back in 2017, about $500 total. All ten have hit end-of-life around the same ten-year mark.

I was ready to serial into one — the board has "RS232" and "serial" printed right next to a port. But before doing that, I bought a new unit instead, the SMICO100, on the idea that if the old Z-Wave daughterboard fits into the new housing, I could skip the reverse-engineering entirely.

Observation

The daughterboard didn't fit. But opening that second unit led to this: across the two units I've now opened — one AA-powered, one 9V-powered — both still had a separate internal battery beyond their primary power source. That didn't add up: why would an externally-powered detector also carry its own internal battery?

Correction

I was wrong about what that cylindrical component is. It's not a battery. It's the carbon monoxide sensor, a Figaro TGS5042, confirmed against Tiny Transistors' teardown of the same First Alert unit, credit to them for the correct ID. It's electrochemical: a sealed tube with air-intake holes, filled with a potassium hydroxide/potassium carbonate electrolyte and a platinum-coated electrode, not a power source.

Screenshot of the Tiny Transistors teardown article correctly identifying the brown cylindrical component as the Figaro TGS5042 carbon monoxide sensor, not a battery

Full archived copy of the source article (PDF): tinytransistors-co-detector.pdf

That means the hypothesis and test below were built on a bad premise. There's no hidden backup battery resetting an operating-time clock. What I actually desoldered and resoldered was the CO sensor itself. I put that unit back into service in the kitchen afterward. Reheating an electrochemical sensor during desoldering can damage the electrolyte or membrane, so I can't trust that unit's CO detection anymore.

Hypothesis (incorrect)

That internal battery isn't there for power — it's there to run a clock, tracking how long the unit has been in operation. That's the actual end-of-life mechanism: not a sensor health check, just an operating-time clock. If that clock's state lives entirely in that battery, disconnecting and reconnecting it should reset the count.

The test (based on the incorrect hypothesis)

I desoldered the battery, let it sit disconnected for 30 seconds, soldered it back onto the same board, and put the unit back into service in the kitchen to see if the end-of-life chirping starts again.