■ FIGURE 2.
the probes are 'shorted' by water, pin 1 is pulled
low, turning on the oscillator. The extremely high
(>1000 Ω) input impedance allows this simple sensor
to work well.
Unfortunately, the current draw was unacceptably
high ( 5-10 mA), even in dry conditions with the
oscillator off. The culprit is the bias on pin 2 of the
first gate. CMOS gates draw very little power only
when the inputs are all connected to ground or Vcc.
A Better Solution
➤ Design #3 — Digging deeper into my junk box, I
came across a CD4047 multivibrator package that
was exactly what I needed. A simplified version of the
internal components of the CD4047 are shown in
Figure 2. In the 'dry' state, NAND gate G2 has one
input low and one high. The low voltage at the gate
of MOSFET Q1 causes it to conduct, pulling one of
the inputs to G2 (and pin 3 of the package) to be
high. In this state, the device draws essentially no
power. None of the internal gates are switching, and
both sides of variable resistor R1 are at 9V.
When water contacts the probes, two things
happen. First, one of the gates of G2 is pulled high,
causing the output of G2 to go low, the output of G3
to go high, and the output of G4 to go low. Second,
the gate of Q1 is also pulled high and Q1 stops
conducting. Since the resting state of capacitor C1
has a 9V difference across it, when the G3 output
goes high the 'top' plate is boosted to 18V. This
voltage sags as R1 slowly drains the current towards 0V.
When C1 drops to 4.5V, G2 switches the output to low,
and the cycle continues. This package has a flip-flop (FF1)
at the output of G2 that divides the frequency in half and
ensures a 50% duty cycle.
A scope view of the signals at pin 10 (top trace 1,
flip-flop output) and at pin 3 (bottom trace 2) is shown in
Figure 3. The vertical scale is 5V/division, and the time
scale is 100 usec/division. The numbers (1 and 2) at the
left side of the scope image are at 0V of each trace.
In the lower trace, the 4.5V switching point for G2 is
The final circuit is shown in Figure 4. It consists of
four discrete parts, one inexpensive CMOS package, and
a probe that was constructed from short pieces of heavy
gauge (#6) solid copper wire. It is easy to construct on
perf board. To adjust R1, the probes are shorted by
touching them and R1 is adjusted for maximum sound
output. I was surprised at the small current draw ( 3 mA)
when the transducer is sounding. The volume is adequate
for inside the house. This CMOS device had been
patiently waiting in my junk box for 25 years, waiting for
a little juice so that it could get busy.
When the probes are dry, the circuit draws less than
my 2,000-count multimeter with a 200 µA scale can
measure. At first, I doubted the accuracy of my meter.
■ FIGURE 4.
Final circuit using
regulator with a 9V battery. I started out by building a
2.5 kHz oscillator from a CD4001 NOR gate package by
linearly biasing the input as shown in Figure 1.
The first NOR gate in this four-gate 14-pin package
is used as a gated oscillator. One of the inputs (pin 2) is
linearly biased by R2, and with capacitor C1, forms a
2.5 kHz oscillator. The remaining three gates are used as
drivers for the piezoelectric transducer. When the probes
(P1) are not in contact with water, pin 1 is high, forcing
the output at pin 3 low, shutting off the oscillator. When
■ FIGURE 3. Oscilloscope view of astable multivibrator in action.
42 June 2009