How the sensors came to be as they are. The present arrangement is described in Sensors.

The tank level sensor

The first tank sensor was a "Connectivity" 5 V DC input, 0-3.3 V DC output analog depth sensor ($54.99 from AliExpress), mounted at the bottom of the tank, with the relationship assumed to be 3.3 V for 3.2 metres of depth: depth = (voltage / 3.3) × 3.2. An RS-485 water depth sensor was also tried, with difficulties. The original tank specification had listed an ultrasonic sensor mounted at the top. The QDY30A 4-20 mA sensor installed in October 2025 replaced them, and the placeholder conversion was replaced by the water-meter calibration of 25 October 2026, which showed the old formula read about 1.2 m too high when the tank was nearly empty - so that the dry, gel and water-saving thresholds could never have triggered.

Temperature sensor defaults

When a temperature sensor was unavailable, AR2 originally substituted believable default values - 31 °C for external, 30 °C for internal and 32 °C for the second external sensor - rather than letting an invalid value propagate. Since November 2026 a failed sensor is sent as 0 with a failure flag, so that it stands out and AR3 can ignore it.

The flow meter

For a long time the DN50 flow meter in the main line did not work correctly (an impeller fault was suspected, needing dismantling), and its calibration constant was a placeholder of about 60 pulses per litre "requiring field calibration". The fault turned out to be noise on the Arduino's 5 V supply; with the meter powered from a clean supply it was calibrated at 16.2 pulses per litre against the house water meter on 25 October 2026 (an intermediate figure of 10.33 had been set in July 2026 against a 290 L/min reference).

The ember detector

The flame sensor was first connected to Arduino 3, and both its digital output and its analog output were tried; both worked well, and the analog pin was used at that stage as slightly more robust, with the code reporting a sustained flame of at least 2 ms. It is now read by AR2 as a digital input. An earlier ember trap design used cloth soaked in hardened paraffin wax as the flammable bed.

The housing for the Pyrotector glowing-ember detector went through several designs before the four-pipe pinwheel box now planned (see Ember detector (planned)). The first mounted a duct on a vane, turning on a boat-seat swivel so that its mouth always faced the wind. It would have worked, but needed a bearing and a slip ring (with a capacitor to ride through brief breaks in the slip-ring contact), and the vane could lag in gusty winds. A second version replaced the standard bends with long-sweep bends, which are gentler on embers but much larger and dearer. Another design used two boxes: an upper "windcatcher" box with a 150 × 60 mm opening in each face, a small angled baffle behind each opening to turn the air downward, and a riveted four-panel aluminium funnel collecting it into a flexible aluminium duct, which ran in an S-bend (to block the light) down to a lower box holding the detector. Fabrication templates were drawn up for it, but it was set aside in favour of the simpler pinwheel box, which has no funnel or baffles to make and carries the embers straight past the lens.

The oil and water pressure wiring

The optocouplers for the oil and water pressure switches were originally wired straight onto the shared switch wires. With the engine running and the oil switch open, the SFS01 box's oil input was dragged down from 4.0 V to 2.1 V, which it read as "no oil pressure", so it re-engaged the starter about every 20 seconds into the running engine (heard as a hiss); the water optocoupler loaded the box's water input in the same way. The fault was diagnosed and fixed on 23-24 October 2026 with the diode-and-pull-up arrangement now described in Sensors.

The shed display

The LCD was at one stage programmed to show four panels of 3 seconds each, including external temperature and water flow. It now shows three pages of 2.5 seconds each.

<< Development history: gel injection | | Development history: the electronics >>      |Table of Contents>


Page last modified on November 08, 2026, at 06:40 AM