If you have read some other articles found at this web site, it should be easy to imagine what the “test site” we mention might be - usually there are a few smart young folks around doing some complicated measurements all night long, they might get sleepy in a pleasantly warmed room while expensive pieces of equipment might get soaked because of a sudden rainstorm perhaps... The circuit we present here has not prevented any really bad misfortune so far, at least we are not informed of such occasions - not because the alarm did not work, but simply because everybody managed to be up to the task and no rain disasters have ever happened.

The most important feature concerning this alarm operation is that is has to react very fast to the very first rain drops that touch the sensor. That is the reason that we opted for detecting rain by a conductive surface and simple electronics that is expected to operate satisfactorily over a wide temperature range (-20 - +40°C), very wide relative humidity range (10 - 100% rH), not to be influenced by wind speed, by relative position of the Sun etc. We remind the readers that the purpose of the alarm is merely to detect the presence of rain outside, while we did not need to measure the actual rate of it. If one wants to measure rain water influx, there is an ingeniously simple and effective technique known under the term “tipping bucket rain gauge” that is often utilised in meteorological stations, and such an apparatus would not be hard at all to construct in amateur circles. But that would not have been an optimal solution to our problem as described previously due to the fact that there is always an unavoidable time delay (sometimes measured in minutes) that passes between the very start of rain and the first valid signal from the tipping bucket. On the other side of the spectrum, there are optical rain sensors able to both detect and measure the severity of rain, that are commonly used in automobiles. To put it simply, a light source (LED) and an optical sensor are put behind the windshield in such a way so that drops of rain disrupt the optical coupling between the two. As this technique is adopted by the highly regulated automotive industry, we can be sure that it works and that it works very well. But the downside is that it is not trivial to implement from ground up in DIY fashion, while commercial devices of this sort are either firmly welded to a windshield (and cannot be cut out as the safety glass would disintegrate) or they output such electrical signals that one would not able to interpret easily, or both. So we decided to spend a few days playing with the little circuit described here.

The device consists of two parts: the outdoor unit (the actual rain sensor) and the indoor unit (light flashing and beeping sound alarm). The previous drawing is the circuit diagram of the sensor. The surface sensitive to water is a piece of PCB onto which the two sets of mutually interleaved conductive lines are formed and exposed to the outdoor air. Gate A is connected to the interleaved lines so that it oscillates at a frequency that depends on the resistance between the lines and the capacitance of C0. As the resistance drops, frequency of the oscillations increases. A pair of 22k resistors connected in series with the sensitive surface limits the maximum frequency, while at the same time they limit the maximum current that might flow through the sensor to merely 0.1mA. This AC signal is fed to what diligent readers of these pages would recognize as “frequency detector circuit” formed out of gates B and C. If the frequency of the signal generated by gate A is above a certain threshold, gate C output turns 0V. This in turn means that outputs of gates D-F are logic high (5V) and that Q1 is put firmly into conduction, turning on the heater formed by connecting more than a dozen 1/4W resistors. Resistors that form the heater are physically more or less evenly spread and placed below the moisture sensitive interleaved lines. What the sensor actually does is that if it finds itself wet, it tries to dry itself by turning on the heater.

The maximum frequency that the oscillator gate A is able to generate is approximately 30kHz. That is in fact the typical value when the sensor is wet, as the resistance of the sensitive surface is then much less than the two 22k resistors connected in series with it. This is high enough to prevent sensor oxidation, while at the same time low enough not to induce electromagnetic interference in nearby devices. The threshold frequency above which gates B and C turn on is set to 7kHz, which practically means that a single drop of rain is enough to trigger the alarm, but that it has to be an actual drop of liquid larger than 1mm across, not some microscopic mist precipitate, a spec of dirt or an insect leg. Note that the start threshold is intentionally designed to be much higher than 50Hz electrical grid frequency in order to prevent triggering of the heater (and consequently of the indoor alarm) due to nearby electrical installations. Once activated, “frequency detector” turns off only after the oscillator frequency falls below 4kHz, which is caused by the hysteretic response of Schmidt-inverter gate C to rectified DC voltage at its input. This in essence requires that, in order for the heater to be turned off, sensor surface has to become appreciably drier than it was at the point at which it caused the heater to turn on. Thus, no frequent turning on and off of the sensor and no false alarms. Our tests show that the minimum duration of the activated alarm is around 15 minutes.

A careful look at the previous circuit diagram reveals that there is a “quiescent heater” that is turned on all the time in it. Its purpose is to raise the temperature of the sensor surface a few degrees above the ambient temperature, so that very wet air itself cannot turn on the alarm if there is no actual precipitation (rain or snow). A side benefit of the quiescent heating is the fact that not only rain but also snow is able to trigger the alarm because snowflakes melt as they touch the sensor. But if one wants the device to alarm its users not merely to rain but as well to heavy fog, then the quiescent heater should be left out.