Translations:Case Studies/43/en

This was my own, little used but worked fine for 2 or 3 years but then stopped working. On switching on, the lights seemed to indicate maybe the water collection tank was full, which it wasn't, and it wouldn't stay on.

• Removed covers, visual inspection showed no signs whatsoever of a problem. Reseated connectors with no effect.
• The water level alarm consisted of a float with a magnet which closed a magnetic reed switch as the float approached the full position. No visible faults, and unable to influence the problem with a second magnet close to the reed switch. Gave up for the time being.
• Determined to have one last go before scrapping it. Followed the reed switch wires through to the logic board and hence to a pin on the microcontroller. Verified with a multimeter that the voltage on the microcontroller pin changed as one would expect on lifting the float.
• Noticed that the silk screen printing on the logic board showed a capacitor which was not fitted.
• Followed the copper traces on the board and found that the capacitor (if fitted) would be wired between another microcontroller pin and ground, with a resistor also connecting the pin to the positive supply.
• Theorised that the microcontroller pin was a reset input. The capacitor (if fitted) would hold the pin in a logic "low" state and hence the microcontroller in a reset state while the capacitor charged, for long enough for the power supplies to stabilise. Without the capacitor, the microcontroller might run with unstable power supply and immediately crash.
• Fitted a capacitor calculated to hold the pin "low" for a few tens of milliseconds.
• Powered the dehumidifier on, and it worked like a dream!

Funnily enough, 2 or 3 years later I had a letter from B&Q. A safety issue had been found with this model of dehumidifier, and would I please bring it back for a full refund. Lucky I'd saved it from the scrap heap - I got my refund and bought a new aluminium ladder with the proceeds!