the years, I've built a few of these to help stop bedwetting.
You can build one too if you have the minimum necessary circuit
Photo 1. The entire alarm fits into a
small plastic box along with a 9 volt battery.
It will be fastened to the user's arm with elastic.
A few days ago, I was asked if I could
make a bedwetting alarm, like those I made years ago. Back then, I used
a Sonalert module (a Piezoelectric alarm) driven by a VMOS FET (VN10KN
from Siliconix) for the alarm. This time, I ran though lots of options,
though my main concerns were minimum standby power consumption and
that the alarm be loud and startling enough to awaken someone from deep
sleep. This one would be of similar construction - the
electronics, battery, and speaker in a box to be worn on the user's
upper arm (See Photo 1), with a cable connecting the electronics
to a pair of
metal snaps attached to the user's pajama bottoms that would conduct
when the user urinated. See Photo 2.
I had decided to use a small 8 Ohm speaker salvaged from a junked cell
phone. A little experimentation with a signal generator showed that it
would be very loud when
driven with only 3 or 4 volts between 800 Hz and 3 kHz. I thought about
ways to drive the speaker: use an LM324 quad op-amp followed by a pair
of transistors, make a transistor multivibrator and have that drive a
transistor buffer, make a couple of oscillators with a hex inverter.
Just about the, the UPS delivery truck pulled up and I took delivery
of a package from Mouser Electronics, which included among its
contents, some LMC555N timer chips that I had ordered in anticipation
of a very different project. This was the solution. The LMC555's can be
set up to oscillate with a 50% duty cycle with only a single resistor
and capacitor -a improvement over the recommended circuit for the old
bipolar NE555, it can operate directly from a 9 volt transistor radio
battery, and can be frequency modulated by injecting a signal into the
control input (pin 5).
Figure 1. The circuit
description from left to right
The circuit shown in Figure 1 is the one that that fell into place.
According to the data sheet, U2's oscillation frequency, F can be found
F = 1/(1.4 R C)
I wanted to run around 2.5 kHz, and chose a 100k resistor, and
found the ideal value of the capacitor (from the formula above):
C = 1/(1.4 R F) = 1/(1.4 x 100K 2500 Hz) = .0028 uf. So, I used the
closest standard value I had on had, which is .0027 uf.
The output of the oscillator drives the complimentary pair made of the
2N2222 and 2N2907. The complimentary pair is necessary because the peak
current into the speaker can reach several hundred milliamps, and the
LMC555 cannot drive it directly. This pair has a tremendous amount of
crossover distortion, so it would not be useful as shown here for
normal audio use, but in this circuit, it is being driven by a
square wave, and that distortion should not have any effect on the
signal, other than to reduce the amplitude a little bit.
A volume control was inserted between the output of the LMC555N (pin 3)
and the bases of the output buffer transistors, because the speaker is
pretty loud and I wanted to give the user the option of turning
it down if desired. This volume control has a fairly high resistance
compared to the input resistance of the emitter followers, but I have a
lot of 10k pots and it is sufficient in this application. If the volume
drifts a bit because of transistor gain drift, that's ok. It will still
work fine. As a rule of thumb, for this sort of volume control, I would
use a pot with a value of less than Zload X Beta, where Zload is the
impedance of the load connected to the emitters and Beta is the current
gain of the transistor expected at the load current. Based on this, the
volume control's maximum resistance should have been about 500 Ohms
(Beta of 50 or more x 10 Ohm load), Given that the highest resistance
on the wiper is when it is in the center, a 500 Ohm maximum output
resistance would be obtained with a 1k pot. But, as I said I have a bag
of 10k pots, so that's what I used.
The constant tone U2 makes may not be enough to wake up a deep sleeper.
If the user did not awaken enough when the alarm first went off, he
could become used to the tone, even though it is loud, and go back to
sleep. To make it harder of the user to go back to sleep, I added a
second oscillator, U1, which runs at about 3 Hz, drives the frequency
modulation input of U2 (pin 5) thorough a 220k resistor. I selected
220k because it seemed appropriately large compared to the string of
100k resistors inside the LMC555, and when I tried it, it sounded just
fine. The result is that when the circuit is on, the speaker is driven
by alternating high and low tones above and below 2500 Hz. I measured
the rate of change, as set by U1, to be 3.3 Hz, by listening to it for
10 seconds and counting 33 changes in frequency.
The circuit has the power applied whenever the battery is installed.
The reset inputs (pin 4) to the LMC555 chips are held low (in the reset
state) by the 1 Meg resistor from the reset inputs to ground.
This means that the oscillators aren't oscillating in the normal state.
When the contacts become wet, they conduct. I figured a resistance of
about 8 Meg or less would cause the chips to come out of reset. A quick
check using some fixed resistors showed that this one triggered when
Megs was placed across the contacts - plenty sensitive, and maybe too
sensitive for some cases. If you want to lower the sensitivity, reduce
the value of the 1 Meg resistor.
A pair of 1N916 diodes keeps the voltage across the "Contacts" input,
such as might be caused by electrostatic discharge, from causing damage
to the reset input of the LMC555's by shunting most of the current to
ground or the battery. A 0.033 uf capacitor across the "Contacts" input
keeps the circuit from responding to stray electrostatic fields such as
may be encountered near an electric blanket.
When the circuit is idling -that is not sounding the alarm, the circuit
only draws about 250 microamps, so it can idle for a long time,
probably months at a time, so there is no on-off switch. I meaured the
average current from the battery at about 35 milliamps when the alarm
was sounding, which is well within the capabilities of the 9 volt
transistor radio battery.
Photo 2. The loops of dark ribbon are handles to allow the snaps to be
separated easily to silence the alarm quickly. They are made from
waxed lacing cord.
The moisture sensor is a pair of clothing snap fasteners purchased from
fabric store, and they are connected to the circuit at the point
labeled "Contacts". They are attached, one on each side of the fabric,
to the user's night garment in a place that would become wet at the
crucial moment. The ones I chose are sew-on nickel plated, which
allowed me to solder the wires to it.
The biggest problem with snaps in the past has been that the wires
break off after a few nights of use. It is therefore, important to be
careful in attaching the wires to the snaps. You might be able to see
from Photo 2 that I looped the wires through the holes in the snaps to
provide a little strain relief for the solder joints. The wires that
connect the snaps to the cable are 22 gauge stranded. I made sure that
solder did not wick up the strands all the way to the insulation so
that the wire could flex a little between the insulation and the solder
joint without breaking. I also tied the wire to the snaps with the
In practice, the snaps can be replaced by other conductive fasteners
such as safety pins or paper clips. I've had success with snaps in the
earlier ones, and recommend them. With any contact arrangement, you
will have to experiment with placement so that they don't accidentally
active, such as in the case of snaps, by wearing a hole in the fabric.
You can send email to me concerning this project at the email address
shown on the HOME
Please see the
liability disclaimer on the HOME
Richard Cappels. All Rights Reserved. http://projects.cappels.org/