Here’s a hypothesis for you: radioactive decay varies over time, possibly with a yearly cycle. [Panteltje] decided to test this hypothesis, and so far has two year’s worth of data to comb over.
Radioactive decay can be easily detected with a photomultiplier tube, but these tubes are sensitive to magnetic fields and cosmic rays that would easily fly through just about any shielding [Pantelje] could come up with. Instead, the radiation in this setup is detected with simple photo detectors, pressed right up against a tritium-filled glass ampoule, a somewhat common lighting solution for fishing lures, watch faces, and compasses.
The experimental setup records the photo detectors, a temperature sensor, and a voltage reference, recording all the data to an EEPROM once an hour. All the important electronics are stuffed into a heatsinked, insulated, light-proof box, while the control electronics reside on a larger board with battery backup, alarm, indicator LEDs, and an RS232 connection.
After one year, [Pantelje] recorded the data and reset the experiment for another year. There are now two years worth of data available, ready for anyone to analyze. Of course, evidence that radioactive decay changes over the course of a few years would turn just about every scientific discipline on its head, so at the very least [Panteltje] has a great record of the output of tritium lights against the expected half-life.