This simple circuit was originally devised as an add-on for the Water Meter project, but can be used as a standalone device or for general experimentation more of which later.

As can be seen from the schematic in Figure 1, the detector is built around a quad 2-input Schmidt trigger NAND gate of which only two gates are used, the two remaining gates have their inputs wired together and taken high. CMOS devices should never have their gates floating and need to be tied high or low, looking at the data sheet for the 4093B the IC consumes less current if spare gates are tied high.

Gate IC1a is wired as a traditional astable oscillator whose frequency is determined by R1, C1 in this case around 1.3KHz.

The output from this gate goes via DC blocking and wave shaping capacitor C2,to one side of the sensor. This capacitor also helps to compensate for stray capacitance caused by the sensor. Under quiescent conditions the square wave output gets as far as terminal X1-1 and goes no further.

When water comes into contact between terminals X1-1 and 2 the circuit is made and the square wave now travels via C3, through the charge pump level restorer C3, D1 and C4,D2. After exiting D2 the square wave is now just a DC high level (+5v), as IC1B is wired as an inverter this causes the output at pin 4 to go low. R5 is used as pin protection to stop excessive current occurring if there is a short or similar for the following uC circuitry.

This resistor can be left out if using either of the output circuits in Figure 2. This just leaves explanations for C4 and R2, these two components help with the sensitivity of the of the device and R2 also ensures that the inputs of IC1b are pulled low in the absence of a signal.

The sensor in its most basic form is just two strips of copper that are shorted out when rain covered by water. Photo 1, shows two examples, one using Veroboard/stripboard and the other proto board.

This particular proto board is ready tinned, if using stripboard then a layer of solder along the tracks will help to stop the copper corroding as shown in the photo. In both cases where the connecting wire is soldered to the tracks some waterproof polymer sealant should be liberally daubed around the connection on both side of the board. This can be the left over sealant from fixing leaking showers and the like. The sensitivity to the type of rain can be altered by experimenting with the distance between the two terminals (the wider the gap the heavier the downpour has to be).

Photo 2 shows two stainless steel dough hooks borrowed from a food mixer that I technically enhanced (in other words it no longer works). These make excellent soil moisture detectors and could be used with the circuit to signal lack of moisture in a plant bed, sensitivity here depending solely on how far apart they are placed and the type of soil.

If not connecting this to a microcontroller, then Figure 2a-c has some suggestions for alternative outputs.

 In 2a, the two spare gates IC1c,d have been paralleled together and connect directly to a LED. 2b shows a connection to a relay, note in this circuit and that of 2c the relay and the switched output may be at a  than that powering IC1. 

IC1 can be powered from 3 – 15V.

There is nothing crucial about this circuit and it can be constructed on stripboard. Photo 3 shows the prototype unit constructed on IC proto board available from Tandy (Radio Shack). The board fits quite nicely into the case used for the water controller, again the casing is left up to you.

Experiments with this basic circuit. An oscilloscope is useful to see waveforms but not essential.

Is trying out any NAND, NOR or INVERTER IC that you may have lying around. They must have Schmidt trigger inputs though. Despite theoretically the 74LS132 ought to work, two samples I had refused to oscillate with the components specified. However mine were from a never built project from 1981 with more recent versions you may have better luck. The 4093 must be the buffered version.

The astable frequency is not that crucial so C1 can be altered, R1 should not be lower than 1K or higher than 1M.

With an oscilloscope watch how the wave form behaves with C2,3 out of circuit or with different values. Values should not be more than 1uF or less than .001uF.

Remove C3,4 and C1, connect pin 3 of IC1 directly to the D1,2 junction. Connect the probe in place of C1.

The probe would need to be altered to two pieces of insulated cable with their far ends waterproofed, the cable will need to be about 12” or so long glued to a plastic substrate – a plastic rule would be fine. Connect one end of the probe to 0v and the other to the input junction of IC1a and R1. You now have a the basis for a capacitance sensor. Due to the impurities of water when the two waterproofed ends are immersed in water there will be an alteration in the capacitance of the probe and the frequency should change. A capacitance meter will show that the capacitance can change by about 30pF if the probes are dipped in water.

If you want the signal output to go high in the alarm condition, then one of the spare gates can be used to invert the output , then you could change the PNP transistor to an NPN equivalent.

The permutations for experimentation is endless.