Theory of Operation
Refer to Figure 1 which shows the schematic. After a
couple false starts, I elected to go with a discrete design
which keeps the unit small, but more importantly, is easy
to run on a unipolar power supply. An electret
microphone is used as the sensor. These are active
devices, with R2 providing the bias from the power supply.
Notice, however, that capacitor C2 blocks the bias
voltage from later stages, while allowing the AC audio
signal to pass. I picked up the parts for this project as
surplus from All Electronics ( www.allelectronics.com),
including the electret microphone. Unbelievably, that set
me back a whopping fifty cents and yet works extremely
well. If you’d like, you could use an external microphone
on a cable with plug, but I went with a soldered-in unit.
The audio signal appearing on the far side of C2 is
quite miniscule at this point —
less than five millivolts — so
we’ll have to preamplify it. Q1
and associated components
carry out that job. The larger
signal is then chained to a
second preamplification stage
configured around Q2.
Both of these stages thus
far (Q1 and Q2) are extremely
primitive, but why open a can
of beans with a stick of
dynamite? They get the job
done, and niceties such as
temperature compensation, flat
response, low noise, and the
like simply aren’t important
when detecting handclaps.
Since the gain is moderately high in both, spurious
oscillation is always a possibility. So, feedback capacitor
C1 is plopped in place to damp it out if it tries to rear its
ugly head. The total gain is around 1,200. Thus, a weak
microphone signal has now become a much beefier three
volts or so.
Sensitivity is dialed in by means of potentiometer R6.
The tamed signal is passed on to the simple half-wave
rectifier. D2 dumps the negative half of the waveform to
ground, while D1 passes the positive portion on to C5. If
you’d like, you can think of this capacitor as a low-pass
filter. Or, if you prefer, as a peak detector. Either way, a
DC voltage proportional to the microphone’s amplitude
envelope is routed to Q3 which more or less switches on
for larger signals.
The transistor may not saturate completely and also
inverts the DC voltage, so we’ll send its output to Q4
which is a true switch now. The output becomes a solid
gate, swinging smartly from 0V to +5V whenever the
amplitude set by R6 exceeds a certain level.
You might think this is all too ingenuous to get the job
done, but keep in mind we’re just trying to sense bursts of
noise and simply want a digital output: on or off. For these
reasons, there is no need to call out the heavy artillery. All
of the usual headaches like response time versus ripple,
resolution, and the like are of no real concern.
Build It in an Evening
The Sonic Sensor is a snap to build and a nice one-night project for DIYers at any level of experience. None
of the parts are hard to find, nor are there any special
construction concerns. It could easily be assembled on a
36 September 2013
■ FIGURE 2.
■ FIGURE 3.